Glossip v. Gross

576 U.S. 863, 135 S. Ct. 2726, 192 L. Ed. 2d 761, 25 Fla. L. Weekly Fed. S 494, 2015 U.S. LEXIS 4255, 83 U.S.L.W. 4656

Opinion

Justice ALITO delivered the opinion of the Court.

Prisoners sentenced to death in the State of Oklahoma filed an action in federal court under Rev. Stat. § 1979, 42 U.S.C. § 1983 , contending that the method of execution now used by the State violates the Eighth Amendment because it creates an unacceptable risk of severe pain. They argue that midazolam, the first drug employed in the State's current three-drug protocol, fails to render a person insensate to pain. After holding an evidentiary hearing, the District Court denied four prisoners' application for a preliminary injunction, finding that they had failed to prove that midazolam is ineffective. The Court of Appeals for the Tenth Circuit affirmed and accepted the District Court's finding of fact regarding midazolam's efficacy.

For two independent reasons, we also affirm. First, the prisoners failed to identify a known and available alternative method of execution that entails a lesser risk of pain, a requirement of all Eighth Amendment method-of-execution claims. See Baze v. Rees, 553 U.S. 35 , 61, 128 S.Ct. 1520 , 170 L.Ed.2d 420 (2008) (plurality opinion). Second, the District Court did not commit clear error when it found that the prisoners failed to establish that Oklahoma's use of a massive dose of midazolam in its execution protocol entails a substantial risk of severe pain.

I

A

The death penalty was an accepted punishment at the time of the adoption of the Constitution and the Bill of Rights. In that era, death sentences were usually carried out by hanging. The Death Penalty in America: Current Controversies 4 (H. Bedau ed. 1997). Hanging remained the standard method of execution through much of the 19th century, but that began to change in the century's later years. See Baze, supra, at 41-42 , 128 S.Ct. 1520 . In the 1880's, the Legislature of the State of New York appointed a commission to find " 'the most humane and practical method known to modern science of carrying into effect the sentence of death in capital cases.' " In re Kemmler, 136 U.S. 436 , 444, 10 S.Ct. 930 , 34 L.Ed. 519 (1890).

*2732 The commission recommended electrocution, and in 1888, the Legislature enacted a law providing for this method of execution. Id., at 444-445 , 10 S.Ct. 930 . In subsequent years, other States followed New York's lead in the " 'belief that electrocution is less painful and more humane than hanging.' " Baze, 553 U.S., at 42 , 128 S.Ct. 1520 (quoting Malloy v. South Carolina, 237 U.S. 180 , 185, 35 S.Ct. 507 , 59 L.Ed. 905 (1915)).

In 1921, the Nevada Legislature adopted another new method of execution, lethal gas, after concluding that this was "the most humane manner known to modern science." State v. Jon, 46 Nev. 418 , 437, 211 P. 676 , 682 (1923). The Nevada Supreme Court rejected the argument that the use of lethal gas was unconstitutional, id., at 435-437 , 211 P., at 681-682 , and other States followed Nevada's lead, see, e.g., Ariz. Const., Art. XXII, § 22 (1933); 1937 Cal. Stats. ch. 172, § 1; 1933 Colo. Sess. Laws ch. 61, § 1; 1955 Md. Laws ch. 625, § 1, p. 1017; 1937 Mo. Laws p. 222, § 1. Nevertheless, hanging and the firing squad were retained in some States, see, e.g., 1961 Del. Laws ch. 309, § 2 (hanging); 1935 Kan. Sess. Laws ch. 155, § 1 (hanging); Utah Code Crim. Proc. § 105-37-16 (1933) (hanging or firing squad), and electrocution remained the predominant method of execution until the 9-year hiatus in executions that ended with our judgment in Gregg v. Georgia, 428 U.S. 153 , 96 S.Ct. 2909 , 49 L.Ed.2d 859 (1976). See Baze, supra, at 42 , 128 S.Ct. 1520 .

After Gregg reaffirmed that the death penalty does not violate the Constitution, some States once again sought a more humane way to carry out death sentences. They eventually adopted lethal injection, which today is "by far the most prevalent method of execution in the United States." Baze, supra, at 42 , 128 S.Ct. 1520 . Oklahoma adopted lethal injection in 1977, see 1977 Okla. Sess. Laws p. 89, and it eventually settled on a protocol that called for the use of three drugs: (1) sodium thiopental, "a fast-acting barbiturate sedative that induces a deep, comalike unconsciousness when given in the amounts used for lethal injection," (2) a paralytic agent, which "inhibits all muscular-skeletal movements and, by paralyzing the diaphragm, stops respiration," and (3) potassium chloride, which "interferes with the electrical signals that stimulate the contractions of the heart, inducing cardiac arrest." Baze, supra, at 44 , 128 S.Ct. 1520 ; see also Brief for Respondents 9. By 2008, at least 30 of the 36 States that used lethal injection employed that particular three-drug protocol. 553 U.S., at 44 , 128 S.Ct. 1520 .

While methods of execution have changed over the years, "[t]his Court has never invalidated a State's chosen procedure for carrying out a sentence of death as the infliction of cruel and unusual punishment." Id., at 48 , 128 S.Ct. 1520 . In Wilkerson v. Utah, 99 U.S. 130 , 134-135, 25 L.Ed. 345 (1879), the Court upheld a sentence of death by firing squad. In In re Kemmler, supra, at 447-449 , 10 S.Ct. 930 , the Court rejected a challenge to the use of the electric chair. And the Court did not retreat from that holding even when presented with a case in which a State's initial attempt to execute a prisoner by electrocution was unsuccessful. Louisiana ex rel. Francis v. Resweber, 329 U.S. 459 , 463-464, 67 S.Ct. 374 , 91 L.Ed. 422 (1947) (plurality opinion). Most recently, in Baze, supra, seven Justices agreed that the three-drug protocol just discussed does not violate the Eighth Amendment.

Our decisions in this area have been animated in part by the recognition that because it is settled that capital punishment is constitutional, "[i]t necessarily follows that there must be a [constitutional]

*2733 means of carrying it out." Id., at 47 , 128 S.Ct. 1520 . And because some risk of pain is inherent in any method of execution, we have held that the Constitution does not require the avoidance of all risk of pain. Ibid. After all, while most humans wish to die a painless death, many do not have that good fortune. Holding that the Eighth Amendment demands the elimination of essentially all risk of pain would effectively outlaw the death penalty altogether.

B

Baze cleared any legal obstacle to use of the most common three-drug protocol that had enabled States to carry out the death penalty in a quick and painless fashion. But a practical obstacle soon emerged, as anti-death-penalty advocates pressured pharmaceutical companies to refuse to supply the drugs used to carry out death sentences. The sole American manufacturer of sodium thiopental, the first drug used in the standard three-drug protocol, was persuaded to cease production of the drug. After suspending domestic production in 2009, the company planned to resume production in Italy. Koppel, Execution Drug Halt Raises Ire of Doctors, Wall Street Journal, Jan. 25, 2011, p. A6. Activists then pressured both the company and the Italian Government to stop the sale of sodium thiopental for use in lethal injections in this country. Bonner, Letter from Europe: Drug Company in Cross Hairs of Death Penalty Opponents, N.Y. Times, Mar. 30, 2011; Koppel, Drug Halt Hinders Executions in the U.S., Wall Street Journal, Jan. 22, 2011, p. A1. That effort proved successful, and in January 2011, the company announced that it would exit the sodium thiopental market entirely. See Hospira, Press Release, Hospira Statement Regarding Pentothal ™ (sodium thiopental) Market Exit (Jan. 21, 2011).

After other efforts to procure sodium thiopental proved unsuccessful, States sought an alternative, and they eventually replaced sodium thiopental with pentobarbital, another barbiturate. In December 2010, Oklahoma became the first State to execute an inmate using pentobarbital. See Reuters, Chicago Tribune, New Drug Mix Used in Oklahoma Execution, Dec. 17 2010, p. 41. That execution occurred without incident, and States gradually shifted to pentobarbital as their supplies of sodium thiopental ran out. It is reported that pentobarbital was used in all of the 43 executions carried out in 2012. The Death Penalty Institute, Execution List 2012, online at www.deathpenalty info.org/execution-list-2012 (all Internet materials as visited June 26, 2015, and available in Clerk of Court's case file). Petitioners concede that pentobarbital, like sodium thiopental, can "reliably induce and maintain a comalike state that renders a person insensate to pain" caused by administration of the second and third drugs in the protocol. Brief for Petitioners 2. And courts across the country have held that the use of pentobarbital in executions does not violate the Eighth Amendment. See, e.g., Jackson v. Danberg, 656 F.3d 157 (C.A.3 2011); Beaty v. Brewer, 649 F.3d 1071 (C.A.9 2011); DeYoung v. Owens, 646 F.3d 1319 (C.A.11 2011); Pavatt v. Jones, 627 F.3d 1336 (C.A.10 2010).

Before long, however, pentobarbital also became unavailable. Anti-death-penalty advocates lobbied the Danish manufacturer of the drug to stop selling it for use in executions. See Bonner, supra . That manufacturer opposed the death penalty and took steps to block the shipment of pentobarbital for use in executions in the United States. Stein, New Obstacle to Death Penalty in U.S., Washington Post, July 3, 2011, p. A4. Oklahoma eventually became unable to acquire the drug through any means. The District Court *2734 below found that both sodium thiopental and pentobarbital are now unavailable to Oklahoma. App. 67-68.

C

Unable to acquire either sodium thiopental or pentobarbital, some States have turned to midazolam, a sedative in the benzodiazepine family of drugs. In October 2013, Florida became the first State to substitute midazolam for pentobarbital as part of a three-drug lethal injection protocol. Fernandez, Executions Stall As States Seek Different Drugs, N.Y. Times, Nov. 9, 2013, p. A1. To date, Florida has conducted 11 executions using that protocol, which calls for midazolam followed by a paralytic agent and potassium chloride. See Brief for State of Florida as Amicus Curiae 2-3; Chavez v. Florida SP Warden, 742 F.3d 1267 , 1269 (C.A.11 2014). In 2014, Oklahoma also substituted midazolam for pentobarbital as part of its three-drug protocol. Oklahoma has already used this three-drug protocol twice: to execute Clayton Lockett in April 2014 and Charles Warner in January 2015. (Warner was one of the four inmates who moved for a preliminary injunction in this case.)

The Lockett execution caused Oklahoma to implement new safety precautions as part of its lethal injection protocol. When Oklahoma executed Lockett, its protocol called for the administration of 100 milligrams of midazolam, as compared to the 500 milligrams that are currently required. On the morning of his execution, Lockett cut himself twice at " 'the bend of the elbow.' " App. 50. That evening, the execution team spent nearly an hour making at least one dozen attempts to establish intravenous (IV) access to Lockett's cardiovascular system, including at his arms and elsewhere on his body. The team eventually believed that it had established intravenous access through Lockett's right femoral vein, and it covered the injection access point with a sheet, in part to preserve Lockett's dignity during the execution. After the team administered the midazolam and a physician determined that Lockett was unconscious, the team next administered the paralytic agent (vecuronium bromide) and most of the potassium chloride. Lockett began to move and speak, at which point the physician lifted the sheet and determined that the IV had "infiltrated," which means that "the IV fluid, rather than entering Lockett's blood stream, had leaked into the tissue surrounding the IV access point." Warner v. Gross, 776 F.3d 721 , 725 (C.A.10 2015) (case below). The execution team stopped administering the remaining potassium chloride and terminated the execution about 33 minutes after the midazolam was first injected. About 10 minutes later, Lockett was pronounced dead.

An investigation into the Lockett execution concluded that "the viability of the IV access point was the single greatest factor that contributed to the difficulty in administering the execution drugs." App. 398. The investigation, which took five months to complete, recommended several changes to Oklahoma's execution protocol, and Oklahoma adopted a new protocol with an effective date of September 30, 2014. That protocol allows the Oklahoma Department of Corrections to choose among four different drug combinations. The option that Oklahoma plans to use to execute petitioners calls for the administration of 500 milligrams of midazolam followed by a paralytic agent and potassium chloride. 1

*2735 The paralytic agent may be pancuronium bromide, vecuronium bromide, or rocuronium bromide, three drugs that, all agree, are functionally equivalent for purposes of this case. The protocol also includes procedural safeguards to help ensure that an inmate remains insensate to any pain caused by the administration of the paralytic agent and potassium chloride. Those safeguards include: (1) the insertion of both a primary and backup IV catheter, (2) procedures to confirm the viability of the IV site, (3) the option to postpone an execution if viable IV sites cannot be established within an hour, (4) a mandatory pause between administration of the first and second drugs, (5) numerous procedures for monitoring the offender's consciousness, including the use of an electrocardiograph and direct observation, and (6) detailed provisions with respect to the training and preparation of the execution team. In January of this year, Oklahoma executed Warner using these revised procedures and the combination of midazolam, a paralytic agent, and potassium chloride.

II

A

In June 2014, after Oklahoma switched from pentobarbital to midazolam and executed Lockett, 21 Oklahoma death row inmates filed an action under 42 U.S.C. § 1983 challenging the State's new lethal injection protocol. The complaint alleged that Oklahoma's use of midazolam violates the Eighth Amendment's prohibition of cruel and unusual punishment.

In November 2014, four of those plaintiffs-Richard Glossip, Benjamin Cole, John Grant, and Warner-filed a motion for a preliminary injunction. All four men had been convicted of murder and sentenced to death by Oklahoma juries. Glossip hired Justin Sneed to kill his employer, Barry Van Treese. Sneed entered a room where Van Treese was sleeping and beat him to death with a baseball bat. See Glossip v. State, 2007 OK CR 12 , 157 P.3d 143 , 147-149. Cole murdered his 9-month-old daughter after she would not stop crying. Cole bent her body backwards until he snapped her spine in half. After the child died, Cole played video games. See Cole v. State, 2007 OK CR 27 , 164 P.3d 1089 , 1092-1093. Grant, while serving terms of imprisonment totaling 130 years, killed Gay Carter, a prison food service supervisor, by pulling her into a mop closet and stabbing her numerous times with a shank. See Grant v. State, 2002 OK CR 36 , 58 P.3d 783 , 789. Warner anally raped and murdered an 11-month-old girl. The child's injuries included two skull fractures, internal brain injuries, two fractures to her jaw, a lacerated liver, and a bruised spleen and lungs. See Warner v. State, 2006 OK CR 40 , 144 P.3d 838 , 856-857.

The Oklahoma Court of Criminal Appeals affirmed the murder conviction and death sentence of each offender. Each of the men then unsuccessfully sought both state postconviction and federal habeas corpus relief. Having exhausted the avenues for challenging their convictions and sentences, they moved for a preliminary injunction against Oklahoma's lethal injection protocol.

B

In December 2014, after discovery, the District Court held a 3-day evidentiary hearing on the preliminary injunction motion. The District Court heard testimony from 17 witnesses and reviewed numerous exhibits. Dr. David Lubarsky, an anesthesiologist, and Dr. Larry Sasich, a doctor of pharmacy, provided expert testimony about midazolam for petitioners, and Dr. Roswell Evans, a doctor of pharmacy, provided expert testimony for respondents.

*2736 After reviewing the evidence, the District Court issued an oral ruling denying the motion for a preliminary injunction. The District Court first rejected petitioners' challenge under Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 , 113 S.Ct. 2786 , 125 L.Ed.2d 469 (1993), to the testimony of Dr. Evans. It concluded that Dr. Evans, the Dean of Auburn University's School of Pharmacy, was well qualified to testify about midazolam's properties and that he offered reliable testimony. The District Court then held that petitioners failed to establish a likelihood of success on the merits of their claim that the use of midazolam violates the Eighth Amendment. The court provided two independent reasons for this conclusion. First, the court held that petitioners failed to identify a known and available method of execution that presented a substantially less severe risk of pain than the method that the State proposed to use. Second, the court found that petitioners failed to prove that Oklahoma's protocol "presents a risk that is 'sure or very likely to cause serious illness and needless suffering,' amounting to 'an objectively intolerable risk of harm.' " App. 96 (quoting Baze, 553 U.S., at 50 , 128 S.Ct. 1520 ). The court emphasized that the Oklahoma protocol featured numerous safeguards, including the establishment of two IV access sites, confirmation of the viability of those sites, and monitoring of the offender's level of consciousness throughout the procedure.

The District Court supported its decision with findings of fact about midazolam. It found that a 500-milligram dose of midazolam "would make it a virtual certainty that any individual will be at a sufficient level of unconsciousness to resist the noxious stimuli which could occur from the application of the second and third drugs." App. 77. Indeed, it found that a 500-milligram dose alone would likely cause death by respiratory arrest within 30 minutes or an hour.

The Court of Appeals for the Tenth Circuit affirmed. 776 F.3d 721 . The Court of Appeals explained that our decision in Baze requires a plaintiff challenging a lethal injection protocol to demonstrate that the risk of severe pain presented by an execution protocol is substantial " 'when compared to the known and available alternatives.' " Id., at 732 (quoting Baze, supra, at 61 , 128 S.Ct. 1520 ). And it agreed with the District Court that petitioners had not identified any such alternative. The Court of Appeals added, however, that this holding was "not outcome-determinative in this case" because petitioners additionally failed to establish that the use of midazolam creates a demonstrated risk of severe pain. 776 F.3d, at 732 . The Court of Appeals found that the District Court did not abuse its discretion by relying on Dr. Evans' testimony, and it concluded that the District Court's factual findings about midazolam were not clearly erroneous. It also held that alleged errors in Dr. Evans' testimony did not render his testimony unreliable or the District Court's findings clearly erroneous.

Oklahoma executed Warner on January 15, 2015, but we subsequently voted to grant review and then stayed the executions of Glossip, Cole, and Grant pending the resolution of this case. 574 U.S. ----, 135 S.Ct. 1173 , 190 L.Ed.2d 929 (2015).

III

"A plaintiff seeking a preliminary injunction must establish that he is likely to succeed on the merits, that he is likely to suffer irreparable harm in the absence of preliminary relief, that the balance of equities tips in his favor, and that an injunction is in the public interest." Winter *2737 v. Natural Resources Defense Council, Inc., 555 U.S. 7 , 20, 129 S.Ct. 365 , 172 L.Ed.2d 249 (2008). The parties agree that this case turns on whether petitioners are able to establish a likelihood of success on the merits.

The Eighth Amendment, made applicable to the States through the Fourteenth Amendment, prohibits the infliction of "cruel and unusual punishments." The controlling opinion in Baze outlined what a prisoner must establish to succeed on an Eighth Amendment method-of-execution claim. Baze involved a challenge by Kentucky death row inmates to that State's three-drug lethal injection protocol of sodium thiopental, pancuronium bromide, and potassium chloride. The inmates conceded that the protocol, if properly administered, would result in a humane and constitutional execution because sodium thiopental would render an inmate oblivious to any pain caused by the second and third drugs. 553 U.S., at 49 , 128 S.Ct. 1520 . But they argued that there was an unacceptable risk that sodium thiopental would not be properly administered. Ibid. The inmates also maintained that a significant risk of harm could be eliminated if Kentucky adopted a one-drug protocol and additional monitoring by trained personnel. Id., at 51 , 128 S.Ct. 1520 .

The controlling opinion in Baze first concluded that prisoners cannot successfully challenge a method of execution unless they establish that the method presents a risk that is " ' sure or very likely to cause serious illness and needless suffering,' and give rise to 'sufficiently imminent dangers.' " Id., at 50 , 128 S.Ct. 1520 (quoting Helling v. McKinney, 509 U.S. 25 , 33, 34-35, 113 S.Ct. 2475 , 125 L.Ed.2d 22 (1993)). To prevail on such a claim, "there must be a 'substantial risk of serious harm,' an 'objectively intolerable risk of harm' that prevents prison officials from pleading that they were 'subjectively blameless for purposes of the Eighth Amendment.' " 553 U.S., at 50 , 128 S.Ct. 1520 (quoting Farmer v. Brennan, 511 U.S. 825 , 846, and n. 9, 114 S.Ct. 1970 , 128 L.Ed.2d 811 (1994)). The controlling opinion also stated that prisoners "cannot successfully challenge a State's method of execution merely by showing a slightly or marginally safer alternative." 553 U.S., at 51 , 128 S.Ct. 1520 . Instead, prisoners must identify an alternative that is "feasible, readily implemented, and in fact significantly reduce[s] a substantial risk of severe pain." Id., at 52 , 128 S.Ct. 1520 .

The controlling opinion summarized the requirements of an Eighth Amendment method-of-execution claim as follows: "A stay of execution may not be granted on grounds such as those asserted here unless the condemned prisoner establishes that the State's lethal injection protocol creates a demonstrated risk of severe pain. [And] [h]e must show that the risk is substantial when compared to the known and available alternatives." Id., at 61 , 128 S.Ct. 1520 . The preliminary injunction posture of the present case thus requires petitioners to establish a likelihood that they can establish both that Oklahoma's lethal injection protocol creates a demonstrated risk of severe pain and that the risk is substantial when compared to the known and available alternatives.

The challenge in Baze failed both because the Kentucky inmates did not show that the risks they identified were substantial and imminent, id., at 56 , 128 S.Ct. 1520 , and because they did not establish the existence of a known and available alternative method of execution that would entail a significantly less severe risk, id., at 57-60 , 128 S.Ct. 1520 . Petitioners' arguments here fail for similar reasons. First, petitioners have not proved that any risk posed by midazolam is substantial *2738 when compared to known and available alternative methods of execution. Second, they have failed to establish that the District Court committed clear error when it found that the use of midazolam will not result in severe pain and suffering. We address each reason in turn.

IV

Our first ground for affirmance is based on petitioners' failure to satisfy their burden of establishing that any risk of harm was substantial when compared to a known and available alternative method of execution. In their amended complaint, petitioners proffered that the State could use sodium thiopental as part of a single-drug protocol. They have since suggested that it might also be constitutional for Oklahoma to use pentobarbital. But the District Court found that both sodium thiopental and pentobarbital are now unavailable to Oklahoma's Department of Corrections. The Court of Appeals affirmed that finding, and it is not clearly erroneous. On the contrary, the record shows that Oklahoma has been unable to procure those drugs despite a good-faith effort to do so.

Petitioners do not seriously contest this factual finding, and they have not identified any available drug or drugs that could be used in place of those that Oklahoma is now unable to obtain. Nor have they shown a risk of pain so great that other acceptable, available methods must be used. Instead, they argue that they need not identify a known and available method of execution that presents less risk. But this argument is inconsistent with the controlling opinion in Baze, 553 U.S., at 61 , 128 S.Ct. 1520 , which imposed a requirement that the Court now follows. 2

Petitioners contend that the requirement to identify an alternative method of execution contravenes our pre- Baze decision in Hill v. McDonough, 547 U.S. 573 , 126 S.Ct. 2096 , 165 L.Ed.2d 44 (2006), but they misread that decision. The portion of the opinion in Hill on which they rely concerned a question of civil procedure, not a substantive Eighth Amendment question. In Hill, the issue was whether a challenge to a method of execution must be brought by means of an application for a writ of habeas corpus or a civil action under § 1983. Id., at 576 , 126 S.Ct. 2096 . We held that a method-of-execution claim must be brought under § 1983 because such a claim does not attack the validity of the prisoner's conviction or death sentence. Id., at 579-580 , 126 S.Ct. 2096 . The United States as amicus curiae argued that we should adopt a special pleading requirement to stop inmates from using § 1983 actions to attack, not just a particular means of execution, but the death penalty itself. To achieve this end, the United States proposed that an inmate asserting a method-of-execution claim should be required to plead an acceptable alternative method of execution. Id., at 582 , 126 S.Ct. 2096 . We rejected that argument because "[s]pecific pleading requirements are mandated *2739 by the Federal Rules of Civil Procedure, and not, as a general rule, through case-by-case determinations of the federal courts." Ibid . Hill thus held that § 1983 alone does not impose a heightened pleading requirement. Baze, on the other hand, addressed the substantive elements of an Eighth Amendment method-of-execution claim, and it made clear that the Eighth Amendment requires a prisoner to plead and prove a known and available alternative. Because petitioners failed to do this, the District Court properly held that they did not establish a likelihood of success on their Eighth Amendment claim.

Readers can judge for themselves how much distance there is between the principal dissent's argument against requiring prisoners to identify an alternative and the view, now announced by Justices BREYER and GINSBURG, that the death penalty is categorically unconstitutional. Post, p. 2759 (BREYER, J., dissenting). The principal dissent goes out of its way to suggest that a State would violate the Eighth Amendment if it used one of the methods of execution employed before the advent of lethal injection. Post, at 2770 - 2772. And the principal dissent makes this suggestion even though the Court held in Wilkerson that this method (the firing squad) is constitutional and even though, in the words of the principal dissent, "there is some reason to think that it is relatively quick and painless." Post, at 2796. Tellingly silent about the methods of execution most commonly used before States switched to lethal injection (the electric chair and gas chamber), the principal dissent implies that it would be unconstitutional to use a method that "could be seen as a devolution to a more primitive era." Ibid. If States cannot return to any of the "more primitive" methods used in the past and if no drug that meets with the principal dissent's approval is available for use in carrying out a death sentence, the logical conclusion is clear. But we have time and again reaffirmed that capital punishment is not per se unconstitutional. See, e.g., Baze, 553 U.S., at 47 , 128 S.Ct. 1520 ; id., at 87-88 , 128 S.Ct. 1520 (SCALIA, J., concurring in judgment); Gregg, 428 U.S., at 187 , 96 S.Ct. 2909 (joint opinion of Stewart, Powell, and Stevens, JJ.); id., at 226 , 96 S.Ct. 2909 (White, J., concurring in judgment); Resweber, 329 U.S., at 464 , 67 S.Ct. 374 ; In re Kemmler, 136 U.S., at 447 , 10 S.Ct. 930 ; Wilkerson, 99 U.S., at 134-135 . We decline to effectively overrule these decisions.

V

We also affirm for a second reason: The District Court did not commit clear error when it found that midazolam is highly likely to render a person unable to feel pain during an execution. We emphasize four points at the outset of our analysis.

First, we review the District Court's factual findings under the deferential "clear error" standard. This standard does not entitle us to overturn a finding "simply because [we are] convinced that [we] would have decided the case differently." Anderson v. Bessemer City, 470 U.S. 564 , 573, 105 S.Ct. 1504 , 84 L.Ed.2d 518 (1985).

Second, petitioners bear the burden of persuasion on this issue. Baze, supra, at 41 , 128 S.Ct. 1520 . Although petitioners expend great effort attacking peripheral aspects of Dr. Evans' testimony, they make little attempt to prove what is critical, i.e., that the evidence they presented to the District Court establishes that the use of midazolam is sure or very likely to result in needless suffering.

Third, numerous courts have concluded that the use of midazolam as the first drug in a three-drug protocol is likely to render an inmate insensate to pain that *2740 might result from administration of the paralytic agent and potassium chloride. See, e.g., 776 F.3d 721 (case below affirming the District Court); Chavez v. Florida SP Warden, 742 F.3d 1267 (affirming the District Court); Banks v. State, 150 So.3d 797 (Fla.2014) (affirming the lower court); Howell v. State, 133 So.3d 511 (Fla.2014) (same); Muhammad v. State, 132 So.3d 176 (Fla.2013) (same). (It is noteworthy that one or both of the two key witnesses in this case-Dr. Lubarsky for petitioners and Dr. Evans for respondents-were witnesses in the Chavez, Howell, and Muhammad cases.) "Where an intermediate court reviews, and affirms, a trial court's factual findings, this Court will not 'lightly overturn' the concurrent findings of the two lower courts." Easley v. Cromartie, 532 U.S. 234 , 242, 121 S.Ct. 1452 , 149 L.Ed.2d 430 (2001). Our review is even more deferential where, as here, multiple trial courts have reached the same finding, and multiple appellate courts have affirmed those findings. Cf. Exxon Co., U.S.A. v. Sofec, Inc., 517 U.S. 830 , 841, 116 S.Ct. 1813 , 135 L.Ed.2d 113 (1996) (explaining that this Court " 'cannot undertake to review concurrent findings of fact by two courts below in the absence of a very obvious and exceptional showing of error' " (quoting Graver Tank & Mfg. Co. v. Linde Air Products Co., 336 U.S. 271 , 275, 69 S.Ct. 535 , 93 L.Ed. 672 (1949))).

Fourth, challenges to lethal injection protocols test the boundaries of the authority and competency of federal courts. Although we must invalidate a lethal injection protocol if it violates the Eighth Amendment, federal courts should not "embroil [themselves] in ongoing scientific controversies beyond their expertise." Baze, supra, at 51 , 128 S.Ct. 1520 . Accordingly, an inmate challenging a protocol bears the burden to show, based on evidence presented to the court, that there is a substantial risk of severe pain.

A

Petitioners attack the District Court's findings of fact on two main grounds. 3 First, they argue that even if midazolam is powerful enough to induce unconsciousness, it is too weak to maintain unconsciousness and insensitivity to pain once the second and third drugs are administered. Second, while conceding that the 500-milligram dose of midazolam is much higher than the normal therapeutic dose, they contend that this fact is irrelevant because midazolam has a "ceiling effect"-that is, at a certain point, an increase in the dose administered will not have any greater effect on the inmate. Neither argument succeeds.

The District Court found that midazolam is capable of placing a person "at a sufficient level of unconsciousness to resist the noxious stimuli which could occur from the application of the second and third drugs." App. 77. This conclusion was not clearly *2741 erroneous. Respondents' expert, Dr. Evans, testified that the proper administration of a 500-milligram dose of midazolam would make it "a virtual certainty" that any individual would be "at a sufficient level of unconsciousness to resist the noxious stimuli which could occur from application of the 2nd and 3rd drugs" used in the Oklahoma protocol. Id., at 302; see also id., at 322. And petitioners' experts acknowledged that they had no contrary scientific proof. See id., at 243-244 (Dr. Sasich stating that the ability of midazolam to render a person insensate to the second and third drugs "has not been subjected to scientific testing"); id ., at 176 (Dr. Lubarsky stating that "there is no scientific literature addressing the use of midazolam as a manner to administer lethal injections in humans").

In an effort to explain this dearth of evidence, Dr. Sasich testified that "[i]t's not my responsibility or the [Food and Drug Administration's] responsibility to prove that the drug doesn't work or is not safe." Tr. of Preliminary Injunction Hearing 357 (Tr.). Instead, he stated, "it's the responsibility of the proponent to show that the drug is safe and effective." Ibid. Dr. Sasich confused the standard imposed on a drug manufacturer seeking approval of a therapeutic drug with the standard that must be borne by a party challenging a State's lethal injection protocol. When a method of execution is authorized under state law, a party contending that this method violates the Eighth Amendment bears the burden of showing that the method creates an unacceptable risk of pain. Here, petitioners' own experts effectively conceded that they lacked evidence to prove their case beyond dispute.

Petitioners attempt to avoid this deficiency by criticizing respondents' expert. They argue that the District Court should not have credited Dr. Evans' testimony because he admitted that his findings were based on " 'extrapolat [ions]' " from studies done about much lower therapeutic doses of midazolam. See Brief for Petitioners 34 (citing Tr. 667-668; emphasis deleted). But because a 500-milligram dose is never administered for a therapeutic purpose, extrapolation was reasonable. And the conclusions of petitioners' experts were also based on extrapolations and assumptions. For example, Dr. Lubarsky relied on "extrapolation of the ceiling effect data." App. 177.

Based on the evidence that the parties presented to the District Court, we must affirm. Testimony from both sides supports the District Court's conclusion that midazolam can render a person insensate to pain. Dr. Evans testified that although midazolam is not an analgesic, it can nonetheless "render the person unconscious and 'insensate' during the remainder of the procedure." Id., at 294. In his discussion about the ceiling effect, Dr. Sasich agreed that as the dose of midazolam increases, it is "expected to produce sedation, amnesia, and finally lack of response to stimuli such as pain (unconsciousness)." Id., at 243. Petitioners argue that midazolam is not powerful enough to keep a person insensate to pain after the administration of the second and third drugs, but Dr. Evans presented creditable testimony to the contrary. See, e.g., Tr. 661 (testifying that a 500-milligram dose of midazolam will induce a coma). 4 Indeed, low doses of midazolam *2742 are sufficient to induce unconsciousness and are even sometimes used as the sole relevant drug in certain medical procedures. Dr. Sasich conceded, for example, that midazolam might be used for medical procedures like colonoscopies and gastroscopies. App. 267-268; see also Brief for Respondents 6-8. 5

Petitioners emphasize that midazolam is not recommended or approved for use as the sole anesthetic during painful surgery, but there are two reasons why this is not dispositive. First, as the District Court found, the 500-milligram dose at issue here "is many times higher than a normal therapeutic dose of midazolam." App. 76. The effect of a small dose of midazolam has minimal probative value about the effect of a 500-milligram dose. Second, the fact that a low dose of midazolam is not the best drug for maintaining unconsciousness during surgery says little about whether a 500-milligram dose of midazolam is constitutionally adequate for purposes of conducting an execution. We recognized this point in Baze, where we concluded that although the medical standard of care might require the use of a blood pressure cuff and an electrocardiogram during surgeries, this does not mean those procedures are required for an execution to pass Eighth Amendment scrutiny. 553 U.S., at 60 , 128 S.Ct. 1520 .

Oklahoma has also adopted important safeguards to ensure that midazolam is properly administered. The District Court emphasized three requirements in particular: The execution team must secure both a primary and backup IV access site, it must confirm the viability of the IV sites, and it must continuously monitor the offender's level of consciousness. The District Court did not commit clear error in concluding that these safeguards help to minimize any risk that might occur in the event that midazolam does not operate as intended. Indeed, we concluded in Baze that many of the safeguards that Oklahoma employs-including the establishment of a primary and backup IV and the presence of personnel to monitor an inmate-help in significantly reducing the risk that an execution protocol will violate the Eighth Amendment. Id., at 55-56 , 128 S.Ct. 1520 . And many other safeguards that Oklahoma has adopted mirror those that the dissent in Baze complained were absent from Kentucky's protocol in that case. For example, the dissent argued that because a consciousness check before injection of the second drug "can reduce a risk of dreadful pain," Kentucky's failure to include that step in its procedure was unconstitutional. Id., at 119 , 128 S.Ct. 1520 (opinion of GINSBURG, J.). The dissent also complained that Kentucky did not monitor the effectiveness of the first drug or pause between injection of the first and second drugs. Id., at 120-121 , 128 S.Ct. 1520 . Oklahoma has accommodated each of those concerns.

B

Petitioners assert that midazolam's "ceiling effect" undermines the District Court's *2743 finding about the effectiveness of the huge dose administered in the Oklahoma protocol. Petitioners argue that midazolam has a "ceiling" above which any increase in dosage produces no effect. As a result, they maintain, it is wrong to assume that a 500-milligram dose has a much greater effect than a therapeutic dose of about 5 milligrams. But the mere fact that midazolam has such a ceiling cannot be dispositive. Dr. Sasich testified that "all drugs essentially have a ceiling effect." Tr. 343. The relevant question here is whether midazolam's ceiling effect occurs below the level of a 500-milligram dose and at a point at which the drug does not have the effect of rendering a person insensate to pain caused by the second and third drugs.

Petitioners provided little probative evidence on this point, and the speculative evidence that they did present to the District Court does not come close to establishing that its factual findings were clearly erroneous. Dr. Sasich stated in his expert report that the literature "indicates" that midazolam has a ceiling effect, but he conceded that he "was unable to determine the midazolam dose for a ceiling effect on unconsciousness because there is no literature in which such testing has been done." App. 243-244. Dr. Lubarsky's report was similar, id., at 171-172 , and the testimony of petitioners' experts at the hearing was no more compelling. Dr. Sasich frankly admitted that he did a "search to try and determine at what dose of midazolam you would get a ceiling effect," but concluded: "I could not find one." Tr. 344. The closest petitioners came was Dr. Lubarsky's suggestion that the ceiling effect occurs "[p]robably after about ... 40 to 50 milligrams," but he added that he had not actually done the relevant calculations, and he admitted: "I can't tell you right now" at what dose the ceiling effect occurs. App. 225. We cannot conclude that the District Court committed clear error in declining to find, based on such speculative evidence, that the ceiling effect negates midazolam's ability to render an inmate insensate to pain caused by the second and third drugs in the protocol.

The principal dissent discusses the ceiling effect at length, but it studiously avoids suggesting that petitioners presented probative evidence about the dose at which the ceiling effect occurs or about whether the effect occurs before a person becomes insensate to pain. The principal dissent avoids these critical issues by suggesting that such evidence is "irrelevant if there is no dose at which the drug can ... render a person 'insensate to pain.' " Post, at 2789. But the District Court heard evidence that the drug can render a person insensate to pain, and not just from Dr. Evans: Dr. Sasich (one of petitioners' own experts) testified that higher doses of midazolam are "expected to produce ... lack of response to stimuli such as pain." App. 243. 6

In their brief, petitioners attempt to deflect attention from their failure of proof regarding midazolam's ceiling effect by criticizing Dr. Evans' testimony. But it was petitioners' burden to establish that midazolam's ceiling occurred at a dosage below the massive 500-milligram dose employed in the Oklahoma protocol and at a point at which the drug failed to render the recipient insensate to pain. They did *2744 not meet that burden, and their criticisms do not undermine Dr. Evans' central point, which the District Court credited, that a properly administered 500-milligram dose of midazolam will render the recipient unable to feel pain.

One of petitioners' criticisms of Dr. Evans' testimony is little more than a quibble about the wording chosen by Dr. Evans at one point in his oral testimony. Petitioners' expert, Dr. Lubarsky, stated in his report that midazolam "increases effective binding of [gamma-aminobutyric acid (GABA) ] to its receptor to induce unconsciousness." 7 App. 172. Dr. Evans' report provided a similar explanation of the way in which midazolam works, see id., at 293-294, and Dr. Lubarsky did not dispute the accuracy of that explanation when he testified at the hearing. Petitioners contend, however, that Dr. Evans erred when he said at the hearing that "[m]idazolam attaches to GABA receptors, inhibiting GABA ." Id., at 312 (emphasis added). Petitioners contend that this statement was incorrect because "far from inhibiting GABA, midazolam facilitates its binding to GABA receptors." Brief for Petitioners 38.

In making this argument, petitioners are simply quarrelling with the words that Dr. Evans used during oral testimony in an effort to explain how midazolam works in terms understandable to a layman. Petitioners do not suggest that the discussion of midazolam in Dr. Evans' expert report was inaccurate, and as for Dr. Evans' passing use of the term "inhibiting," Dr. Lubarsky's own expert report states that GABA's " inhibition of brain activity is accentuated by midazolam." App. 232 (emphasis added). Dr. Evans' oral use of the word "inhibiting"-particularly in light of his written testimony-does not invalidate the District Court's decision to rely on his testimony.

Petitioners also point to an apparent conflict between Dr. Evans' testimony and a declaration by Dr. Lubarsky (submitted after the District Court ruled) regarding the biological process that produces midazolam's ceiling effect. But even if Dr. Lubarsky's declaration is correct, it is largely beside the point. What matters for present purposes is the dosage at which the ceiling effect kicks in, not the biological process that produces the effect. And Dr. Lubarsky's declaration does not render the District Court's findings clearly erroneous with respect to that critical issue.

C

Petitioners' remaining arguments about midazolam all lack merit. First, we are not persuaded by petitioners' argument that Dr. Evans' testimony should have been rejected because of some of the sources listed in his report. Petitioners criticize two of the "selected references" that Dr. Evans cited in his expert report: the Web site drugs.com and a material safety data sheet (MSDS) about midazolam. Petitioners' argument is more of a Daubert challenge to Dr. Evans' testimony than an argument that the District Court's findings were clearly erroneous. The District Court concluded that Dr. Evans was "well-qualified to give the expert testimony that he gave" and that "his testimony was the product of reliable principles and methods reliably applied to the facts of this case." App. 75-76. To the extent that the reliability of Dr. Evans' testimony is even before us, the District Court's conclusion that his testimony was based on reliable sources is reviewed under the deferential *2745 "abuse-of-discretion" standard. General Elec. Co. v. Joiner, 522 U.S. 136 , 142-143, 118 S.Ct. 512 , 139 L.Ed.2d 508 (1997). Dr. Evans relied on multiple sources and his own expertise, and his testimony may not be disqualified simply because one source (drugs.com) warns that it " 'is not intended for medical advice' " and another (the MSDS) states that its information is provided " 'without any warranty, express or implied, regarding its correctness.' " Brief for Petitioners 36. Medical journals that both parties rely upon typically contain similar disclaimers. See, e.g., Anesthesiology, Terms and Conditions of Use, online at http://anesthesiology. pubs.asahq.org/ss/terms.aspx ("None of the information on this Site shall be used to diagnose or treat any health problem or disease"). Dr. Lubarsky-petitioners' own expert-relied on an MSDS to argue that midazolam has a ceiling effect. And petitioners do not identify any incorrect statements from drugs.com on which Dr. Evans relied. In fact, although Dr. Sasich submitted a declaration to the Court of Appeals criticizing Dr. Evans' reference to drugs.com, that declaration does not identify a single fact from that site's discussion of midazolam that was materially inaccurate.

Second, petitioners argue that Dr. Evans' expert report contained a mathematical error, but we find this argument insignificant. Dr. Evans stated in his expert report that the lowest dose of midazolam resulting in human deaths, according to an MSDS, is 0.071 mg/kg delivered intravenously. App. 294. Dr. Lubarsky agreed with this statement. Specifically, he testified that fatalities have occurred in doses ranging from 0.04 to 0.07 mg/kg, and he stated that Dr. Evans' testimony to that effect was "a true statement" (though he added those fatalities occurred among the elderly). Id., at 217 . We do not understand petitioners to dispute the testimony of Dr. Evans and their own expert that 0.071 mg/kg is a potentially fatal dose of midazolam. Instead, they make much of the fact that the MSDS attached to Dr. Evans' report apparently contained a typographical error and reported the lowest toxic dose as 71 mg/kg. That Dr. Evans did not repeat that incorrect figure but instead reported the correct dose supports rather than undermines his testimony. In any event, the alleged error in the MSDS is irrelevant because the District Court expressly stated that it did not rely on the figure in the MSDS. See id., at 75.

Third, petitioners argue that there is no consensus among the States regarding midazolam's efficacy because only four States (Oklahoma, Arizona, Florida, and Ohio) have used midazolam as part of an execution. Petitioners rely on the plurality's statement in Baze that "it is difficult to regard a practice as 'objectively intolerable' when it is in fact widely tolerated," and the plurality's emphasis on the fact that 36 States had adopted lethal injection and 30 States used the particular three-drug protocol at issue in that case. 553 U.S., at 53 , 128 S.Ct. 1520 . But while the near-universal use of the particular protocol at issue in Baze supported our conclusion that this protocol did not violate the Eighth Amendment, we did not say that the converse was true, i.e., that other protocols or methods of execution are of doubtful constitutionality. That argument, if accepted, would hamper the adoption of new and potentially more humane methods of execution and would prevent States from adapting to changes in the availability of suitable drugs.

Fourth, petitioners argue that difficulties with Oklahoma's execution of Lockett and Arizona's July 2014 execution of Joseph Wood establish that midazolam is sure or very likely to cause serious pain. We are not persuaded. Aside from the *2746 Lockett execution, 12 other executions have been conducted using the three-drug protocol at issue here, and those appear to have been conducted without any significant problems. See Brief for Respondents 32; Brief for State of Florida as Amicus Curiae 1. Moreover, Lockett was administered only 100 milligrams of midazolam, and Oklahoma's investigation into that execution concluded that the difficulties were due primarily to the execution team's inability to obtain an IV access site. And the Wood execution did not involve the protocol at issue here. Wood did not receive a single dose of 500 milligrams of midazolam; instead, he received fifteen 50-milligram doses over the span of two hours. 8 Brief for Respondents 12, n. 9. And Arizona used a different two-drug protocol that paired midazolam with hydromorphone, a drug that is not at issue in this case. Ibid. When all of the circumstances are considered, the Lockett and Wood executions have little probative value for present purposes.

Finally, we find it appropriate to respond to the principal dissent's groundless suggestion that our decision is tantamount to allowing prisoners to be "drawn and quartered, slowly tortured to death, or actually burned at the stake." Post, at 2795. That is simply not true, and the principal dissent's resort to this outlandish rhetoric reveals the weakness of its legal arguments.

VI

For these reasons, the judgment of the Court of Appeals for the Tenth Circuit is affirmed.

It is so ordered.

Justice SCALIA, with whom Justice THOMAS joins, concurring.

I join the opinion of the Court, and write to respond to Justice BREYER's plea for judicial abolition of the death penalty.

Welcome to Groundhog Day. The scene is familiar: Petitioners, sentenced to die for the crimes they committed (including, in the case of one petitioner since put to death, raping and murdering an 11-month-old baby), come before this Court asking us to nullify their sentences as "cruel and unusual" under the Eighth Amendment. They rely on this provision because it is the only provision they can rely on. They were charged by a sovereign State with murder. They were afforded counsel and tried before a jury of their peers-tried twice, once to determine *2747 whether they were guilty and once to determine whether death was the appropriate sentence. They were duly convicted and sentenced. They were granted the right to appeal and to seek postconviction relief, first in state and then in federal court. And now, acknowledging that their convictions are unassailable, they ask us for clemency, as though clemency were ours to give.

The response is also familiar: A vocal minority of the Court, waving over their heads a ream of the most recent abolitionist studies (a superabundant genre) as though they have discovered the lost folios of Shakespeare, insist that now, at long last, the death penalty must be abolished for good. Mind you, not once in the history of the American Republic has this Court ever suggested the death penalty is categorically impermissible. The reason is obvious: It is impossible to hold unconstitutional that which the Constitution explicitly contemplates . The Fifth Amendment provides that "[n]o person shall be held to answer for a capital ... crime, unless on a presentment or indictment of a Grand Jury," and that no person shall be "deprived of life ... without due process of law." Nevertheless, today Justice BREYER takes on the role of the abolitionists in this long-running drama, arguing that the text of the Constitution and two centuries of history must yield to his "20 years of experience on this Court," and inviting full briefing on the continued permissibility of capital punishment, post, at 2781 (dissenting opinion).

Historically, the Eighth Amendment was understood to bar only those punishments that added " 'terror, pain, or disgrace' " to an otherwise permissible capital sentence. Baze v. Rees, 553 U.S. 35 , 96, 128 S.Ct. 1520 , 170 L.Ed.2d 420 (2008) (THOMAS, J., concurring in judgment). Rather than bother with this troubling detail, Justice BREYER elects to contort the constitutional text. Redefining "cruel" to mean "unreliable," "arbitrary," or causing "excessive delays," and "unusual" to include a "decline in use," he proceeds to offer up a white paper devoid of any meaningful legal argument.

Even accepting Justice BREYER's rewriting of the Eighth Amendment, his argument is full of internal contradictions and (it must be said) gobbledy-gook. He says that the death penalty is cruel because it is unreliable; but it is convictions, not punishments, that are unreliable. Moreover, the "pressure on police, prosecutors, and jurors to secure a conviction," which he claims increases the risk of wrongful convictions in capital cases, flows from the nature of the crime, not the punishment that follows its commission. Post, at 2757 - 2758. Justice BREYER acknowledges as much: "[T]he crimes at issue in capital cases are typically horrendous murders, and thus accompanied by intense community pressure." Ibid. That same pressure would exist, and the same risk of wrongful convictions, if horrendous death-penalty cases were converted into equally horrendous life-without-parole cases. The reality is that any innocent defendant is infinitely better off appealing a death sentence than a sentence of life imprisonment. (Which, again, Justice BREYER acknowledges: "[C]ourts (or State Governors) are 130 times more likely to exonerate a defendant where a death sentence is at issue," post, at 2757.) The capital convict will obtain endless legal assistance from the abolition lobby (and legal favoritism from abolitionist judges), while the lifer languishes unnoticed behind bars.

Justice BREYER next says that the death penalty is cruel because it is arbitrary. To prove this point, he points to a study of 205 cases that "measured the 'egregiousness' of the murderer's conduct"

*2748 with "a system of metrics," and then "compared the egregiousness of the conduct of the 9 defendants sentenced to death with the egregiousness of the conduct of defendants in the remaining 196 cases [who were not sentenced to death]," post, at 2760. If only Aristotle, Aquinas, and Hume knew that moral philosophy could be so neatly distilled into a pocket-sized, vade mecum "system of metrics." Of course it cannot: Egregiousness is a moral judgment susceptible of few hard-and-fast rules. More importantly, egregiousness of the crime is only one of several factors that render a punishment condign-culpability, rehabilitative potential, and the need for deterrence also are relevant. That is why this Court has required an individualized consideration of all mitigating circumstances, rather than formulaic application of some egregiousness test.

It is because these questions are contextual and admit of no easy answers that we rely on juries to make judgments about the people and crimes before them. The fact that these judgments may vary across cases is an inevitable consequence of the jury trial, that cornerstone of Anglo-American judicial procedure. But when a punishment is authorized by law-if you kill you are subject to death-the fact that some defendants receive mercy from their jury no more renders the underlying punishment "cruel" than does the fact that some guilty individuals are never apprehended, are never tried, are acquitted, or are pardoned.

Justice BREYER's third reason that the death penalty is cruel is that it entails delay, thereby (1) subjecting inmates to long periods on death row and (2) undermining the penological justifications of the death penalty. The first point is nonsense. Life without parole is an even lengthier period than the wait on death row; and if the objection is that death row is a more confining environment, the solution should be modifying the environment rather than abolishing the death penalty. As for the argument that delay undermines the penological rationales for the death penalty: In insisting that "the major alternative to capital punishment-namely, life in prison without possibility of parole-also incapacitates," post, at 2767, Justice BREYER apparently forgets that one of the plaintiffs in this very case was already in prison when he committed the murder that landed him on death row. Justice BREYER further asserts that "whatever interest in retribution might be served by the death penalty as currently administered, that interest can be served almost as well by a sentence of life in prison without parole," post, at 2769. My goodness. If he thinks the death penalty not much more harsh (and hence not much more retributive), why is he so keen to get rid of it? With all due respect, whether the death penalty and life imprisonment constitute more-or-less equivalent retribution is a question far above the judiciary's pay grade. Perhaps Justice BREYER is more forgiving-or more enlightened-than those who, like Kant, believe that death is the only just punishment for taking a life. I would not presume to tell parents whose life has been forever altered by the brutal murder of a child that life imprisonment is punishment enough.

And finally, Justice BREYER speculates that it does not "seem likely" that the death penalty has a "significant" deterrent effect. Post, at 2768. It seems very likely to me, and there are statistical studies that say so. See, e.g., Zimmerman, State Executions, Deterrence, and the Incidence of Murder, 7 J. Applied Econ. 163, 166 (2004) ("[I]t is estimated that each state execution deters approximately fourteen murders per year on average"); Dezhbakhsh, Rubin, & Shepherd, *2749 Does Capital Punishment Have a Deterrent Effect? New Evidence from Postmoratorium Panel Data, 5 Am. L. & Econ. Rev. 344 (2003) ("[E]ach execution results, on average, in eighteen fewer murders" per year); Sunstein & Vermeule, Is Capital Punishment Morally Required? Acts, Omissions, and Life-Life Tradeoffs, 58 Stan. L. Rev. 703 , 713 (2005) ("All in all, the recent evidence of a deterrent effect from capital punishment seems impressive, especially in light of its 'apparent power and unanimity' "). But we federal judges live in a world apart from the vast majority of Americans. After work, we retire to homes in placid suburbia or to high-rise co-ops with guards at the door. We are not confronted with the threat of violence that is ever present in many Americans' everyday lives. The suggestion that the incremental deterrent effect of capital punishment does not seem "significant" reflects, it seems to me, a let-them-eat-cake obliviousness to the needs of others. Let the People decide how much incremental deterrence is appropriate.

Of course, this delay is a problem of the Court's own making. As Justice BREYER concedes, for more than 160 years, capital sentences were carried out in an average of two years or less. Post, at 2764. But by 2014, he tells us, it took an average of 18 years to carry out a death sentence. Id., at 2764 - 2765. What happened in the intervening years? Nothing other than the proliferation of labyrinthine restrictions on capital punishment, promulgated by this Court under an interpretation of the Eighth Amendment that empowered it to divine "the evolving standards of decency that mark the progress of a maturing society," Trop v. Dulles, 356 U.S. 86 , 101, 78 S.Ct. 590 , 2 L.Ed.2d 630 (1958) (plurality opinion)-a task for which we are eminently ill suited. Indeed, for the past two decades, Justice BREYER has been the Drum Major in this parade. His invocation of the resultant delay as grounds for abolishing the death penalty calls to mind the man sentenced to death for killing his parents, who pleads for mercy on the ground that he is an orphan. Amplifying the surrealism of his argument, Justice BREYER uses the fact that many States have abandoned capital punishment-have abandoned it precisely because of the costs those suspect decisions have imposed-to conclude that it is now "unusual." Post, at 2772 - 2776. (A caution to the reader: Do not use the creative arithmetic that Justice BREYER employs in counting the number of States that use the death penalty when you prepare your next tax return; outside the world of our Eighth Amendment abolitionist-inspired jurisprudence, it will be regarded as more misrepresentation than math.)

If we were to travel down the path that Justice BREYER sets out for us and once again consider the constitutionality of the death penalty, I would ask that counsel also brief whether our cases that have abandoned the historical understanding of the Eighth Amendment, beginning with Trop, should be overruled. That case has caused more mischief to our jurisprudence, to our federal system, and to our society than any other that comes to mind. Justice BREYER's dissent is the living refutation of Trop 's assumption that this Court has the capacity to recognize "evolving standards of decency." Time and again, the People have voted to exact the death penalty as punishment for the most serious of crimes. Time and again, this Court has upheld that decision. And time and again, a vocal minority of this Court has insisted that things have "changed radically," post, at 2755, and has sought to replace the judgments of the People with their own standards of decency.

Capital punishment presents moral questions that philosophers, theologians, *2750 and statesmen have grappled with for millennia. The Framers of our Constitution disagreed bitterly on the matter. For that reason, they handled it the same way they handled many other controversial issues: they left it to the People to decide. By arrogating to himself the power to overturn that decision, Justice BREYER does not just reject the death penalty, he rejects the Enlightenment.

Justice THOMAS, with whom Justice SCALIA joins, concurring.

I agree with the Court that petitioners' Eighth Amendment claim fails. That claim has no foundation in the Eighth Amendment, which prohibits only those "method[s] of execution" that are "deliberately designed to inflict pain." Baze v. Rees, 553 U.S. 35 , 94, 128 S.Ct. 1520 , 170 L.Ed.2d 420 (2008) (THOMAS, J., concurring in judgment). Because petitioners make no allegation that Oklahoma adopted its lethal injection protocol "to add elements of terror, pain, or disgrace to the death penalty," they have no valid claim. Id., at 107 , 128 S.Ct. 1520 . That should have been the end of this case, but our precedents have predictably transformed the federal courts "into boards of inquiry charged with determining the 'best practices' for executions," id., at 101 , 128 S.Ct. 1520 (internal quotation marks omitted), necessitating the painstaking factual inquiry the Court undertakes today. Although I continue to believe that the broader interpretation of the Eighth Amendment advanced in the plurality opinion in Baze is erroneous, I join the Court's opinion in full because it correctly explains why petitioners' claim fails even under that controlling opinion.

I write separately to respond to Justice BREYER's dissent questioning the constitutionality of the death penalty generally. No more need be said about the constitutional arguments on which Justice BREYER relies, as my colleagues and I have elsewhere refuted them. 1 But Justice *2751 BREYER's assertion, post, at 2760, that the death penalty in this country has fallen short of the aspiration that capital punishment be reserved for the "worst of the worst"-a notion itself based on an implicit proportionality principle that has long been discredited, see Harmelin v. Michigan, 501 U.S. 957 , 966, 111 S.Ct. 2680 , 115 L.Ed.2d 836 (1991) (opinion of SCALIA, J.)-merits further comment. His conclusion is based on an analysis that itself provides a powerful case against enforcing an imaginary constitutional rule against "arbitrariness."

The thrust of Justice Breyer's argument is that empirical studies performed by death penalty abolitionists reveal that the assignment of death sentences does not necessarily correspond to the "egregiousness" of the crimes, but instead appears to be correlated to "arbitrary" factors, such as the locality in which the crime was committed. Relying on these studies to determine the constitutionality of the death penalty fails to respect the values implicit in the Constitution's allocation of decisionmaking in this context. The Donohue study, on which Justice BREYER relies most heavily, measured the "egregiousness" (or "deathworthiness") of murders by asking lawyers to identify the legal grounds for aggravation in each case, and by asking law students to evaluate written summaries of the murders and assign "egregiousness" scores based on a rubric designed to capture and standardize their moral judgments. Donohue, An Empirical Evaluation of the Connecticut Death Penalty System Since 1973, Are There Unlawful Racial, Gender, and Geographic Disparities? 11 J. of Empirical Legal Studies 637, 644-645 (2014). This exercise in some ways approximates the function performed by jurors, but there is at least one critical difference: The law students make their moral judgments based on written summaries-they do not sit through hours, days, or weeks of evidence detailing the crime; they do not have an opportunity to assess the credibility of witnesses, to see the remorse of the defendant, to feel the impact of the crime on the victim's family; they do not bear the burden of deciding the fate of another human being; and they are not drawn from the community whose sense of security and justice may have been torn asunder by an act of callous disregard for human life. They are like appellate judges and justices, reviewing only a paper record of each side's case for life or death.

There is a reason the choice between life and death, within legal limits, is left to the jurors and judges who sit through the trial, and not to legal elites (or law students). 2 That reason is memorialized not once, but twice, in our Constitution: Article III guarantees that "[t]he Trial of all *2752 Crimes, except in cases of Impeachment, shall be by Jury" and that "such Trial shall be held in the State where the said Crimes shall have been committed." Art. III, § 2, cl. 3. And the Sixth Amendment promises that "[i]n all criminal prosecutions, the accused shall enjoy the right to a ... trial, by an impartial jury of the State and district wherein the crime shall have been committed." Those provisions ensure that capital defendants are given the option to be sentenced by a jury of their peers who, collectively, are better situated to make the moral judgment between life and death than are the products of contemporary American law schools.

It should come as no surprise, then, that the primary explanation a regression analysis revealed for the gap between the egregiousness scores and the actual sentences was not the race or sex of the offender or victim, but the locality in which the crime was committed. Donohue, supra, at 640; see also post, at 2761 (BREYER, J., dissenting). What is more surprising is that Justice BREYER considers this factor to be evidence of arbitrariness. See ibid. The constitutional provisions just quoted, which place such decisions in the hands of jurors and trial courts located where "the crime shall have been committed," seem deliberately designed to introduce that factor.

In any event, the results of these studies are inherently unreliable because they purport to control for egregiousness by quantifying moral depravity in a process that is itself arbitrary, not to mention dehumanizing. One such study's explanation of how the author assigned "depravity points" to identify the "worst of the worst" murderers proves the point well. McCord, Lightning Still Strikes, 71 Brooklyn L. Rev. 797, 833-834 (2005). Each aggravating factor received a point value based on the "blameworth[iness]" of the action associated with it. Id., at 830. Killing a prison guard, for instance, earned a defendant three "depravity points" because it improved the case for complete incapacitation, while killing a police officer merited only two, because, "considered dispassionately," such acts do "not seem be a sine qua non of the worst criminals." Id., at 834-836. (Do not worry, the author reassures us, "many killers of police officers accrue depravity points in other ways that clearly put them among the worst criminals." Id., at 836.) Killing a child under the age of 12 was worth two depravity points, because such an act "seems particularly heartless," but killing someone over the age of 70 earned the murderer only one, for although "elderly victims tug at our hearts," they do so "less" than children "because the promise of a long life is less." Id., at 836, 838. Killing to make a political statement was worth three depravity points; killing out of racial hatred, only two. Id., at 835, 837. It goes on, but this small sample of the moral judgments on which this study rested shows just how unsuitable this evidence is to serve as a basis for a judicial decision declaring unconstitutional a punishment duly enacted in more than 30 States, and by the Federal Government.

We owe victims more than this sort of pseudoscientific assessment of their lives. It is bad enough to tell a mother that her child's murder is not "worthy" of society's ultimate expression of moral condemnation. But to do so based on cardboard stereotypes or cold mathematical calculations is beyond my comprehension. In my decades on the Court, I have not seen a capital crime that could not be considered sufficiently "blameworthy" to merit a death sentence (even when genuine constitutional errors justified a vacatur of that sentence). 3

*2753 A small sample of the applications for a stay of execution that have come before the Court this Term alone proves my point. Mark Christeson was due to be executed in October 2014 for his role in the murder of Susan Brouk and her young children, Adrian and Kyle. After raping Ms. Brouk at gunpoint, he and his accomplice drove the family to a remote pond, where Christeson cut Ms. Brouk's throat with a bone knife. State v. Christeson, 50 S.W.3d 251 , 257-258 (Mo.2001). Although bleeding profusely, she stayed alive long enough to tell her children she loved them and to watch as Christeson murdered them-her son, by cutting his throat twice and drowning him; her daughter, by pressing down on her throat until she suffocated. Ibid. Christeson and his accomplice then threw Ms. Brouk-alive but barely breathing-into the pond to drown on top of her dead children. Ibid. This Court granted him a stay of execution. Christeson v. Roper, 574 U.S. ----, 135 S.Ct. 14 , 190 L.Ed.2d 322 (2014). Lisa Ann Coleman was not so lucky. She was executed on September 17, 2014, for murdering her girlfriend's son, 9-year-old Davontae Williams, by slowly starving him to death. Coleman v. State, 2009 WL 4696064 , *1 (Tex.Crim.App., Dec. 9, 2009). When he died, Davontae had over 250 distinct injuries-including cigarette burns and ligature marks-on his 36-pound frame. Id., at *2. Infections from untreated wounds contributed to his other cause of death: pneumonia. Id., at *1-*2. And Johnny Shane Kormondy, who met his end on January 15, 2015, did so after he and his two accomplices invaded the home of a married couple, took turns raping the wife and forcing her to perform oral sex at gunpoint-at one point, doing both simultaneously-and then put a bullet in her husband's head during the final rape. Kormondy v. Secretary, Fla. Dept. of Corrections, 688 F.3d 1244 , 1247-1248 (C.A.11 2012).

Some of our most "egregious" cases have been those in which we have granted relief based on an unfounded Eighth Amendment claim. For example, we have granted relief in a number of egregious cases based on this Court's decision in Atkins v. Virginia, 536 U.S. 304 , 122 S.Ct. 2242 , 153 L.Ed.2d 335 (2002), exempting certain "mentally retarded" offenders from the death penalty. Last Term, the Court granted relief to a man who kidnaped, beat, raped, and murdered a 21-year-old pregnant newlywed, Karol Hurst, also murdering her unborn child, and then, on the same day, murdered a sheriff's deputy *2754 acting in the line of duty. Hall v. Florida, 572 U.S. ----, ----, 134 S.Ct. 1986 , 1990, 188 L.Ed.2d 1007 (2014). And in Atkins itself, the Court granted relief to a man who carjacked Eric Michael Nesbitt, forced him to withdraw money from a bank, drove him to a secluded area, and then shot him multiple times before leaving him to bleed to death. Atkins v. Commonwealth, 257 Va. 160 , 166-167, 510 S.E.2d 445 , 449-450 (1999).

The Court has also misinterpreted the Eighth Amendment to grant relief in egregious cases involving rape. In Kennedy v. Louisiana, 554 U.S. 407 , 128 S.Ct. 2641 , 171 L.Ed.2d 525 (2008), the Court granted relief to a man who had been sentenced to death for raping his 8-year-old stepdaughter. The rape was so violent that it "separated her cervix from the back of her vagina, causing her rectum to protrude into the vaginal structure," and tore her "entire perineum ... from the posterior fourchette to the anus." Id., at 414 , 128 S.Ct. 2641 . The evidence indicated that the petitioner spent at least an hour and half attempting to destroy the evidence of his crime before seeking emergency assistance, even as his stepdaughter bled profusely from her injuries. Id., at 415 , 128 S.Ct. 2641 . And in Coker v. Georgia, 433 U.S. 584 , 97 S.Ct. 2861 , 53 L.Ed.2d 982 (1977) (plurality opinion), the Court granted relief to a petitioner who had escaped from prison, broken into the home of a young married couple and their newborn, forced the wife to bind her husband, gagged her husband with her underwear, raped her (even after being told that she was recovering from a recent childbirth), and then kidnaped her after threatening her husband, Coker v. State, 234 Ga. 555 , 556-557, 216 S.E.2d 782 , 786-787 (1975). In each case, the Court crafted an Eighth Amendment right to be free from execution for the crime of rape-whether it be of an adult, Coker, 433 U.S., at 592 , 97 S.Ct. 2861 , or a child, Kennedy, supra, at 413 , 128 S.Ct. 2641 .

The Court's recent decision finding that the Eighth Amendment prohibits the execution of those who committed their crimes as juveniles is no different. See Roper v. Simmons, 543 U.S. 551 , 125 S.Ct. 1183 , 161 L.Ed.2d 1 (2005). Although the Court had rejected the claim less than two decades earlier, Stanford v. Kentucky, 492 U.S. 361 , 109 S.Ct. 2969 , 106 L.Ed.2d 306 (1989), it decided to revisit the issue for a petitioner who had slain his victim because "he wanted to murder someone" and believed he could "get away with it" because he was a few months shy of his 18th birthday. 543 U.S., at 556 , 125 S.Ct. 1183 . His randomly chosen victim was Shirley Crook, whom he and his friends kidnaped in the middle of the night, bound with duct tape and electrical wire, and threw off a bridge to drown in the river below. Id., at 556-557 , 125 S.Ct. 1183 . The State of Alabama's brief in that case warned the Court that its decision would free from death row a number of killers who had been sentenced for crimes committed as juveniles. Brief for State of Alabama et al. as Amici Curiae in Roper v. Simmons, O.T. 2014, No. 03-633. Mark Duke, for example, murdered his father for refusing to loan him a truck, and his father's girlfriend and her two young daughters because he wanted no witnesses to the crime. Id., at 4. He shot his father and his father's girlfriend pointblank in the face as they pleaded for their lives. Id., at 5-6. He then tracked the girls down in their hiding places and slit their throats, leaving them alive for several minutes as they drowned in their own blood. Id., at 6-7.

Whatever one's views on the permissibility or wisdom of the death penalty, I doubt anyone would disagree that each of these crimes was egregious enough to merit the severest condemnation that society has to *2755 offer. The only constitutional problem with the fact that these criminals were spared that condemnation, while others were not, is that their amnesty came in the form of unfounded claims. Arbitrariness has nothing to do with it. 4 To the extent that we are ill at ease with these disparate outcomes, it seems to me that the best solution is for the Court to stop making up Eighth Amendment claims in its ceaseless quest to end the death penalty through undemocratic means.

Justice BREYER, with whom Justice GINSBURG joins, dissenting.

For the reasons stated in Justice SOTOMAYOR's opinion, I dissent from the Court's holding. But rather than try to patch up the death penalty's legal wounds one at a time, I would ask for full briefing on a more basic question: whether the death penalty violates the Constitution.

The relevant legal standard is the standard set forth in the Eighth Amendment. The Constitution there forbids the "inflict[ion]" of "cruel and unusual punishments." Amdt. 8. The Court has recognized that a "claim that punishment is excessive is judged not by the standards that prevailed in 1685 when Lord Jeffreys presided over the 'Bloody Assizes' or when the Bill of Rights was adopted, but rather by those that currently prevail." Atkins v. Virginia, 536 U.S. 304 , 311, 122 S.Ct. 2242 , 153 L.Ed.2d 335 (2002). Indeed, the Constitution prohibits various gruesome punishments that were common in Blackstone's day. See 4 W. Blackstone, Commentaries on the Laws of England 369-370 (1769) (listing mutilation and dismembering, among other punishments).

Nearly 40 years ago, this Court upheld the death penalty under statutes that, in the Court's view, contained safeguards sufficient to ensure that the penalty would be applied reliably and not arbitrarily. See Gregg v. Georgia, 428 U.S. 153 , 187, 96 S.Ct. 2909 , 49 L.Ed.2d 859 (1976) (joint opinion of Stewart, Powell, and Stevens, JJ.); Proffitt v. Florida, 428 U.S. 242 , 247, 96 S.Ct. 2960 , 49 L.Ed.2d 913 (1976) (joint opinion of Stewart, Powell, and Stevens, JJ.); Jurek v. Texas, 428 U.S. 262 , 268, 96 S.Ct. 2950 , 49 L.Ed.2d 929 (1976) (joint opinion of Stewart, Powell, and Stevens, JJ.); but cf. Woodson v. North Carolina, 428 U.S. 280 , 303, 96 S.Ct. 2978 , 49 L.Ed.2d 944 (1976) (plurality opinion) (striking down mandatory death penalty); Roberts v. Louisiana, 428 U.S. 325 , 331, 96 S.Ct. 3001 , 49 L.Ed.2d 974 (1976) (plurality opinion) (similar). The circumstances and the evidence of the death penalty's application have changed radically since then. Given those changes, I believe that it is now time to reopen the question.

In 1976, the Court thought that the constitutional infirmities in the death penalty could be healed; the Court in effect delegated significant responsibility to the States to develop procedures that would protect against those constitutional problems. Almost 40 years of studies, surveys, and experience strongly indicate, however, that this effort has failed. Today's administration of the death penalty involves three fundamental constitutional defects:

*2756 (1) serious unreliability, (2) arbitrariness in application, and (3) unconscionably long delays that undermine the death penalty's penological purpose. Perhaps as a result, (4) most places within the United States have abandoned its use.

I shall describe each of these considerations, emphasizing changes that have occurred during the past four decades. For it is those changes, taken together with my own 20 years of experience on this Court, that lead me to believe that the death penalty, in and of itself, now likely constitutes a legally prohibited "cruel and unusual punishmen[t]." U.S. Const., Amdt. 8.

I

"Cruel"-Lack of Reliability

This Court has specified that the finality of death creates a "qualitative difference" between the death penalty and other punishments (including life in prison). Woodson, 428 U.S., at 305 , 96 S.Ct. 2978 (plurality opinion). That "qualitative difference" creates "a corresponding difference in the need for reliability in the determination that death is the appropriate punishment in a specific case." Ibid. There is increasing evidence, however, that the death penalty as now applied lacks that requisite reliability. Cf. Kansas v. Marsh, 548 U.S. 163 , 207-211, 126 S.Ct. 2516 , 165 L.Ed.2d 429 (2006) (Souter, J., dissenting) (DNA exonerations constitute "a new body of fact" when considering the constitutionality of capital punishment).

For one thing, despite the difficulty of investigating the circumstances surrounding an execution for a crime that took place long ago, researchers have found convincing evidence that, in the past three decades, innocent people have been executed. See, e.g., Liebman, Fatal Injustice; Carlos DeLuna's Execution Shows That a Faster, Cheaper Death Penalty is a Dangerous Idea, L.A. Times, June 1, 2012, p. A19 (describing results of a 4-year investigation, later published as The Wrong Carlos: Anatomy of a Wrongful Execution (2014), that led its authors to conclude that Carlos DeLuna, sentenced to death and executed in 1989, six years after his arrest in Texas for stabbing a single mother to death in a convenience store, was innocent); Grann, Trial By Fire: Did Texas Execute An Innocent Man? The New Yorker, Sept. 7, 2009, p. 42 (describing evidence that Cameron Todd Willingham was convicted, and ultimately executed in 2004, for the apparently motiveless murder of his three children as the result of invalid scientific analysis of the scene of the house fire that killed his children). See also, e.g., Press Release: Gov. Ritter Grants Posthumous Pardon in Case Dating Back to 1930s, Jan. 7, 2011, p. 1 (Colorado Governor granted full and unconditional posthumous pardon to Joe Arridy, a man with an IQ of 46 who was executed in 1936, because, according to the Governor, "an overwhelming body of evidence indicates the 23-year-old Arridy was innocent, including false and coerced confessions, the likelihood that Arridy was not in Pueblo at the time of the killing, and an admission of guilt by someone else"); R. Warden, Wilkie Collins's The Dead Alive: The Novel, the Case, and Wrongful Convictions 157-158 (2005) (in 1987, Nebraska Governor Bob Kerrey pardoned William Jackson Marion, who had been executed a century earlier for the murder of John Cameron, a man who later turned up alive; the alleged victim, Cameron, had gone to Mexico to avoid a shotgun wedding).

For another, the evidence that the death penalty has been wrongly imposed (whether or not it was carried out), is striking. As of 2002, this Court used the word "disturbing" to describe the number of instances in which individuals had been sentenced *2757 to death but later exonerated. At that time, there was evidence of approximately 60 exonerations in capital cases. Atkins, 536 U.S., at 320, n. 25 , 122 S.Ct. 2242 ; National Registry of Exonerations, online at http://www.law.umich.edu/special/exoneration/Pages/about.aspx (all Internet materials as visited June 25, 2015, and available in Clerk of Court's case file). (I use "exoneration" to refer to relief from all legal consequences of a capital conviction through a decision by a prosecutor, a Governor or a court, after new evidence of the defendant's innocence was discovered.) Since 2002, the number of exonerations in capital cases has risen to 115. Ibid. ; National Registry of Exonerations, Exonerations in the United States, 1989-2012, pp. 6-7 (2012) (Exonerations 2012 Report) (defining exoneration); accord, Death Penalty Information Center (DPIC), Innocence: List of Those Freed from Death Row, online at http://www.deathpenaltyinfo.org/innocence-and-death-penalty (DPIC Innocence List) (calculating, under a slightly different definition of exoneration, the number of exonerations since 1973 as 154). Last year, in 2014, six death row inmates were exonerated based on actual innocence. All had been imprisoned for more than 30 years (and one for almost 40 years) at the time of their exonerations. National Registry of Exonerations, Exonerations in 2014, p. 2 (2015).

The stories of three of the men exonerated within the last year are illustrative. DNA evidence showed that Henry Lee McCollum did not commit the rape and murder for which he had been sentenced to death. Katz & Eckholm, DNA Evidence Clears Two Men in 1983 Murder, N.Y. Times, Sept. 3, 2014, p. A1. Last Term, this Court ordered that Anthony Ray Hinton, who had been convicted of murder, receive further hearings in state court; he was exonerated earlier this year because the forensic evidence used against him was flawed. Hinton v. Alabama, 571 U.S. ----, 134 S.Ct. 1081 , 188 L.Ed.2d 1 (2014) ( per curiam ); Blinder, Alabama Man on Death Row for Three Decades Is Freed as State's Case Erodes, N.Y. Times, Apr. 4, 2014, p. A11. And when Glenn Ford, also convicted of murder, was exonerated, the prosecutor admitted that even "[a]t the time this case was tried there was evidence that would have cleared Glenn Ford." Stroud, Lead Prosecutor Apologizes for Role in Sending Man to Death Row, Shreveport Times, Mar. 27, 2015. All three of these men spent 30 years on death row before being exonerated. I return to these examples infra .

Furthermore, exonerations occur far more frequently where capital convictions, rather than ordinary criminal convictions, are at issue. Researchers have calculated that courts (or State Governors) are 130 times more likely to exonerate a defendant where a death sentence is at issue. They are nine times more likely to exonerate where a capital murder, rather than a noncapital murder, is at issue. Exonerations 2012 Report 15-16, and nn. 24-26.

Why is that so? To some degree, it must be because the law that governs capital cases is more complex. To some degree, it must reflect the fact that courts scrutinize capital cases more closely. But, to some degree, it likely also reflects a greater likelihood of an initial wrongful conviction . How could that be so? In the view of researchers who have conducted these studies, it could be so because the crimes at issue in capital cases are typically horrendous murders, and thus accompanied by intense community pressure on police, prosecutors, and jurors to secure a conviction. This pressure creates a greater likelihood of convicting the wrong person. See Gross, Jacoby, Matheson, Montgomery, & Patil, *2758 Exonerations in the United States 1989 Through 2003, 95 J. Crim. L. & C. 523, 531-533 (2005); Gross & O'Brien, Frequency and Predictors of False Conviction: Why We Know So Little, and New Data on Capital Cases, 5 J. Empirical L. Studies 927, 956-957 (2008) (noting that, in comparing those who were exonerated from death row to other capital defendants who were not so exonerated, the initial police investigations tended to be shorter for those exonerated); see also B. Garrett, Convicting the Innocent: Where Criminal Prosecutions Go Wrong (2011) (discussing other common causes of wrongful convictions generally including false confessions, mistaken eyewitness testimony, untruthful jailhouse informants, and ineffective defense counsel).

In the case of Cameron Todd Willingham, for example, who (as noted earlier) was executed despite likely innocence, the State Bar of Texas recently filed formal misconduct charges against the lead prosecutor for his actions-actions that may have contributed to Willingham's conviction. Possley, Prosecutor Accused of Misconduct in Death Penalty Case, Washington Post, Mar. 19, 2015, p. A3. And in Glenn Ford's case, the prosecutor admitted that he was partly responsible for Ford's wrongful conviction, issuing a public apology to Ford and explaining that, at the time of Ford's conviction, he was "not as interested in justice as [he] was in winning." Stroud, supra .

Other factors may also play a role. One is the practice of death-qualification; no one can serve on a capital jury who is not willing to impose the death penalty. See Rozelle, The Principled Executioner: Capital Juries' Bias and the Benefits of True Bifurcation, 38 Ariz. S.L.J. 769, 772-793, 807 (2006) (summarizing research and concluding that "[f]or over fifty years, empirical investigation has demonstrated that death qualification skews juries toward guilt and death"); Note, Mandatory Voir Dire Questions in Capital Cases: A Potential Solution to the Biases of Death Qualification, 10 Roger Williams Univ. L. Rev. 211, 214-223 (2004) (similar).

Another is the more general problem of flawed forensic testimony. See Garrett, supra, at 7. The Federal Bureau of Investigation (FBI), for example, recently found that flawed microscopic hair analysis was used in 33 of 35 capital cases under review; 9 of the 33 had already been executed. FBI, National Press Releases, FBI Testimony on Microscopic Hair Analysis Contained Errors in at Least 90 Percent of Cases in Ongoing Review, Apr. 20, 2015. See also Hsu, FBI Admits Errors at Trials: False Matches on Crime-Scene Hair, Washington Post, Apr. 19, 2015, p. A1 (in the District of Columbia, which does not have the death penalty, five of seven defendants in cases with flawed hair analysis testimony were eventually exonerated).

In light of these and other factors, researchers estimate that about 4% of those sentenced to death are actually innocent. See Gross, O'Brien, Hu, & Kennedy, Rate of False Conviction of Criminal Defendants Who Are Sentenced to Death, 111 Proceeding of the National Academy of Sciences 7230 (2014) (full-scale study of all death sentences from 1973 through 2004 estimating that 4.1% of those sentenced to death are actually innocent); Risinger, Innocents Convicted: An Empirically Justified Factual Wrongful Conviction Rate, 97 J. Crim. L. & C. 761 (2007) (examination of DNA exonerations in death penalty cases for murder-rapes between 1982 and 1989 suggesting an analogous rate of between 3.3% and 5%).

Finally, if we expand our definition of "exoneration" (which we limited to errors suggesting the defendant was actually innocent) and thereby also categorize as "erroneous" instances in which courts failed *2759 to follow legally required procedures, the numbers soar. Between 1973 and 1995, courts identified prejudicial errors in 68% of the capital cases before them. Gelman, Liebman, West, & Kiss, A Broken System: The Persistent Patterns of Reversals of Death Sentences in the United States, 1 J. Empirical L. Studies 209, 217 (2004). State courts on direct and postconviction review overturned 47% of the sentences they reviewed. Id., at 232. Federal courts, reviewing capital cases in habeas corpus proceedings, found error in 40% of those cases. Ibid.

This research and these figures are likely controversial. Full briefing would allow us to scrutinize them with more care. But, at a minimum, they suggest a serious problem of reliability. They suggest that there are too many instances in which courts sentence defendants to death without complying with the necessary procedures; and they suggest that, in a significant number of cases, the death sentence is imposed on a person who did not commit the crime. See Earley, A Pink Cadillac, An IQ of 63, and A Fourteen-Year-Old from South Carolina: Why I Can No Longer Support the Death Penalty, 49 U. Rich. L. Rev. 811 , 813 (2015) ("I have come to the conclusion that the death penalty is based on a false utopian premise. That false premise is that we have had, do have, will have 100% accuracy in death penalty convictions and executions"); Earley, I Oversaw 36 Executions. Even Death Penalty Supporters Can Push for Change, Guardian, May 12, 2014 (Earley presided over 36 executions as Virginia Attorney General from 1998-2001); but see ante, at 2747 - 2748 (SCALIA, J., concurring) (apparently finding no special constitutional problem arising from the fact that the execution of an innocent person is irreversible). Unlike 40 years ago, we now have plausible evidence of unreliability that (perhaps due to DNA evidence) is stronger than the evidence we had before. In sum, there is significantly more research-based evidence today indicating that courts sentence to death individuals who may well be actually innocent or whose convictions (in the law's view) do not warrant the death penalty's application.

II

"Cruel"-Arbitrariness

The arbitrary imposition of punishment is the antithesis of the rule of law. For that reason, Justice Potter Stewart (who supplied critical votes for the holdings in Furman v. Georgia, 408 U.S. 238 , 92 S.Ct. 2726 , 33 L.Ed.2d 346 (1972) ( per curiam ), and Gregg ) found the death penalty unconstitutional as administered in 1972:

"These death sentences are cruel and unusual in the same way that being struck by lightning is cruel and unusual. For, of all the people convicted of [death-eligible crimes], many just as reprehensible as these, the[se] petitioners are among a capriciously selected random handful upon which the sentence of death has in fact been imposed." Furman, 408 U.S., at 309-310 , 92 S.Ct. 2726 (concurring opinion).

See also id., at 310 , 92 S.Ct. 2726 ("[T]he Eighth and Fourteenth Amendments cannot tolerate the infliction of a sentence of death under legal systems that permit this unique penalty to be so wantonly and so freakishly imposed"); id., at 313 , 92 S.Ct. 2726 (White, J., concurring) ("[T]he death penalty is exacted with great infrequency even for the most atrocious crimes and ... there is no meaningful basis for distinguishing the few cases in which it is imposed from the many cases in which it is not").

When the death penalty was reinstated in 1976, this Court acknowledged that the death penalty is (and would be) unconstitutional *2760 if "inflicted in an arbitrary and capricious manner." Gregg, 428 U.S., at 188 , 96 S.Ct. 2909 (joint opinion of Stewart, Powell, and Stevens, JJ.); see also id., at 189 , 96 S.Ct. 2909 ("[W]here discretion is afforded a sentencing body on a matter so grave as the determination of whether a human life should be taken or spared, that discretion must be suitably directed and limited so as to minimize the risk of wholly arbitrary and capricious action"); Godfrey v. Georgia, 446 U.S. 420 , 428, 100 S.Ct. 1759 , 64 L.Ed.2d 398 (1980) (plurality opinion) (similar).

The Court has consequently sought to make the application of the death penalty less arbitrary by restricting its use to those whom Justice Souter called " 'the worst of the worst.' " Kansas v. Marsh, 548 U.S., at 206 , 126 S.Ct. 2516 (dissenting opinion); see also Roper v. Simmons, 543 U.S. 551 , 568, 125 S.Ct. 1183 , 161 L.Ed.2d 1 (2005) ("Capital punishment must be limited to those offenders who commit a narrow category of the most serious crimes and whose extreme culpability makes them the most deserving of execution" (internal quotation marks omitted)); Kennedy v. Louisiana, 554 U.S. 407 , 420, 128 S.Ct. 2641 , 171 L.Ed.2d 525 (2008) (citing Roper, supra, at 568, 125 S.Ct. 1183 ).

Despite the Gregg Court's hope for fair administration of the death penalty, 40 years of further experience make it increasingly clear that the death penalty is imposed arbitrarily, i.e., without the "reasonable consistency" legally necessary to reconcile its use with the Constitution's commands. Eddings v. Oklahoma, 455 U.S. 104 , 112, 102 S.Ct. 869 , 71 L.Ed.2d 1 (1982).

Thorough studies of death penalty sentences support this conclusion. A recent study, for example, examined all death penalty sentences imposed between 1973 and 2007 in Connecticut, a State that abolished the death penalty in 2012. Donohue, An Empirical Evaluation of the Connecticut Death Penalty System Since 1973: Are There Unlawful Racial, Gender, and Geographic Disparities? 11 J. Empirical Legal Studies 637 (2014). The study reviewed treatment of all homicide defendants. It found 205 instances in which Connecticut law made the defendant eligible for a death sentence. Id., at 641-643. Courts imposed a death sentence in 12 of these 205 cases, of which 9 were sustained on appeal. Id., at 641. The study then measured the "egregiousness" of the murderer's conduct in those 9 cases, developing a system of metrics designed to do so. Id., at 643-645. It then compared the egregiousness of the conduct of the 9 defendants sentenced to death with the egregiousness of the conduct of defendants in the remaining 196 cases (those in which the defendant, though found guilty of a death-eligible offense, was ultimately not sentenced to death). Application of the studies' metrics made clear that only 1 of those 9 defendants was indeed the "worst of the worst" (or was, at least, within the 15% considered most "egregious"). The remaining eight were not. Their behavior was no worse than the behavior of at least 33 and as many as 170 other defendants (out of a total pool of 205) who had not been sentenced to death. Id., at 678-679.

Such studies indicate that the factors that most clearly ought to affect application of the death penalty-namely, comparative egregiousness of the crime-often do not. Other studies show that circumstances that ought not to affect application of the death penalty, such as race, gender, or geography, often do .

Numerous studies, for example, have concluded that individuals accused of murdering white victims, as opposed to black or other minority victims, are more likely to receive the death penalty. See GAO, *2761 Report to the Senate and House Committees on the Judiciary: Death Penalty Sentencing 5 (GAO/GGD-90-57, 1990) (82% of the 28 studies conducted between 1972 and 1990 found that race of victim influences capital murder charge or death sentence, a "finding ... remarkably consistent across data sets, states, data collection methods, and analytic techniques"); Shatz & Dalton, Challenging the Death Penalty with Statistics: Furman, McCleskey, and a Single County Case Study, 34 Cardozo L. Rev. 1227 , 1245-1251 (2013) (same conclusion drawn from 20 plus studies conducted between 1990 and 2013).

Fewer, but still many, studies have found that the gender of the defendant or the gender of the victim makes a not-otherwise-warranted difference. Id., at 1251-1253 (citing many studies).

Geography also plays an important role in determining who is sentenced to death. See id., at 1253-1256 . And that is not simply because some States permit the death penalty while others do not. Rather within a death penalty State, the imposition of the death penalty heavily depends on the county in which a defendant is tried. Smith, The Geography of the Death Penalty and its Ramifications, 92 B. U. L. Rev. 227 , 231-232 (2012) (hereinafter Smith); see also Donohue, supra, at 673 ("[T]he single most important influence from 1973-2007 explaining whether a death-eligible defendant [in Connecticut] would be sentenced to death was whether the crime occurred in Waterbury [County]"). Between 2004 and 2009, for example, just 29 counties (fewer than 1% of counties in the country) accounted for approximately half of all death sentences imposed nationwide. Smith 233. And in 2012, just 59 counties (fewer than 2% of counties in the country) accounted for all death sentences imposed nationwide. DPIC, The 2% Death Penalty: How A Minority of Counties Produce Most Death Cases At Enormous Costs to All 9 (Oct. 2013).

What accounts for this county-by-county disparity? Some studies indicate that the disparity reflects the decisionmaking authority, the legal discretion, and ultimately the power of the local prosecutor. See, e.g., Goelzhauser, Prosecutorial Discretion Under Resource Constraints: Budget Allocations and Local Death-Charging Decisions, 96 Judicature 161, 162-163 (2013); Barnes, Sloss, & Thaman, Place Matters (Most): An Empirical Study of Prosecutorial Decision-Making in Death-Eligible Cases, 51 Ariz. L. Rev. 305 (2009) (analyzing Missouri); Donohue, An Empirical Evaluation of the Connecticut Death Penalty System, at 681 (Connecticut); Marceau, Kamin, & Foglia, Death Eligibility in Colorado: Many Are Called, Few Are Chosen, 84 U. Colo. L. Rev. 1069 (2013) (Colorado); Shatz & Dalton, supra, at 1260-1261 (Alameda County).

Others suggest that the availability of resources for defense counsel (or the lack thereof) helps explain geographical differences. See, e.g., Smith 258-265 (counties with higher death-sentencing rates tend to have weaker public defense programs); Liebman & Clarke, Minority Practice, Majority's Burden: The Death Penalty Today, 9 Ohio S. J. Crim. L. 255, 274 (2011) (hereinafter Liebman & Clarke) (similar); see generally Bright, Counsel for the Poor: The Death Sentence Not for the Worst Crime but for the Worst Lawyer, 103 Yale L. J. 1835 (1994).

Still others indicate that the racial composition of and distribution within a county plays an important role. See, e.g., Levinson, Smith, & Young, Devaluing Death: An Empirical Study of Implicit Racial Bias on Jury-Eligible Citizens in Six Death Penalty States, 89 N.Y.U. L. Rev. 513 , 533-536 (2014) (summarizing research on *2762 this point); see also Shatz & Dalton, supra, at 1275 (describing research finding that death-sentencing rates were lowest in counties with the highest nonwhite population); cf. Cohen & Smith, The Racial Geography of the Federal Death Penalty, 85 Wash. L. Rev. 425 (2010) (arguing that the federal death penalty is sought disproportionately where the federal district, from which the jury will be drawn, has a dramatic racial difference from the county in which the federal crime occurred).

Finally, some studies suggest that political pressures, including pressures on judges who must stand for election, can make a difference. See Woodward v. Alabama, 571 U.S. ----, ----, 134 S.Ct. 405 , 408, 187 L.Ed.2d 449 (2013) (SOTOMAYOR, J., dissenting from denial of certiorari) (noting that empirical evidence suggests that, when Alabama judges reverse jury recommendations, these "judges, who are elected in partisan proceedings, appear to have succumbed to electoral pressures"); Harris v. Alabama, 513 U.S. 504 , 519, 115 S.Ct. 1031 , 130 L.Ed.2d 1004 (1995) (Stevens, J., dissenting) (similar); Gelman, 1 J. Empirical L. Studies, at 247 (elected state judges are less likely to reverse flawed verdicts in capital cases in small towns than in larger communities).

Thus, whether one looks at research indicating that irrelevant or improper factors-such as race, gender, local geography, and resources- do significantly determine who receives the death penalty, or whether one looks at research indicating that proper factors-such as "egregiousness"-do not determine who receives the death penalty, the legal conclusion must be the same: The research strongly suggests that the death penalty is imposed arbitrarily.

Justice THOMAS catalogues the tragic details of various capital cases, ante, at 2752 - 2755 (concurring opinion), but this misses my point. Every murder is tragic, but unless we return to the mandatory death penalty struck down in Woodson, 428 U.S., at 304-305 , 96 S.Ct. 2978 , the constitutionality of capital punishment rests on its limited application to the worst of the worst, supra, at 2759 - 2760. And this extensive body of evidence suggests that it is not so limited.

Four decades ago, the Court believed it possible to interpret the Eighth Amendment in ways that would significantly limit the arbitrary application of the death sentence. See Gregg, 428 U.S., at 195 , 96 S.Ct. 2909 (joint opinion of Stewart, Powell, and Stevens, JJ.) ("[T]he concerns expressed in Furman that the penalty of death not be imposed in an arbitrary or capricious manner can be met"). But that no longer seems likely.

The Constitution does not prohibit the use of prosecutorial discretion. Id., at 199 , and n. 50, 96 S.Ct. 2909 (joint opinion of Stewart, Powell, and Stevens, JJ.); McCleskey v. Kemp, 481 U.S. 279 , 307-308, and n. 28, 311-312, 107 S.Ct. 1756 , 95 L.Ed.2d 262 (1987). It has not proved possible to increase capital defense funding significantly. Smith, The Supreme Court and the Politics of Death, 94 Va. L. Rev. 283 , 355 (2008) ("Capital defenders are notoriously underfunded, particularly in states ... that lead the nation in executions"); American Bar Assn. (ABA) Guidelines for the Appointment and Performance of Defense Counsel in Death Penalty Cases, Guideline 9.1, Commentary (rev. ed. Feb. 2003), in 31 Hofstra L. Rev. 913 , 985 (2003) ( "[C]ompensation of attorneys for death penalty representation remains notoriously inadequate"). And courts cannot easily inquire into judicial motivation. See, e.g., Harris, supra .

Moreover, racial and gender biases may, unfortunately, reflect deeply rooted community biases (conscious or unconscious), *2763 which, despite their legal irrelevance, may affect a jury's evaluation of mitigating evidence, see Callins v. Collins, 510 U.S. 1141 , 1153, 114 S.Ct. 1127 , 127 L.Ed.2d 435 (1994) (Blackmun, J., dissenting from denial of certiorari) ("Perhaps it should not be surprising that the biases and prejudices that infect society generally would influence the determination of who is sentenced to death"). Nevertheless, it remains the jury's task to make the individualized assessment of whether the defendant's mitigation evidence entitles him to mercy. See, e.g., Penry v. Lynaugh, 492 U.S. 302 , 319, 109 S.Ct. 2934 , 106 L.Ed.2d 256 (1989); Lockett v. Ohio, 438 U.S. 586 , 604-605, 98 S.Ct. 2954 , 57 L.Ed.2d 973 (1978) (opinion of Burger, C.J.); Woodson, 428 U.S., at 304-305 , 96 S.Ct. 2978 (plurality opinion).

Finally, since this Court held that comparative proportionality review is not constitutionally required, Pulley v. Harris, 465 U.S. 37 , 104 S.Ct. 871 , 79 L.Ed.2d 29 (1984), it seems unlikely that appeals can prevent the arbitrariness I have described. See Kaufman-Osborn, Capital Punishment, Proportionality Review, and Claims of Fairness (with Lessons from Washington State), 79 Wash. L. Rev. 775 , 791-792 (2004) (after Pulley, many States repealed their statutes requiring comparative proportionality review, and most state high courts "reduced proportionality review to a perfunctory exercise" (internal quotation marks omitted)).

The studies bear out my own view, reached after considering thousands of death penalty cases and last-minute petitions over the course of more than 20 years. I see discrepancies for which I can find no rational explanations. Cf. Godfrey, 446 U.S., at 433 , 100 S.Ct. 1759 (plurality opinion) ("There is no principled way to distinguish this case, in which the death penalty was imposed, from the many cases in which it was not"). Why does one defendant who committed a single-victim murder receive the death penalty (due to aggravators of a prior felony conviction and an after-the-fact robbery), while another defendant does not, despite having kidnapped, raped, and murdered a young mother while leaving her infant baby to die at the scene of the crime. Compare State v. Badgett, 361 N.C. 234 , 644 S.E.2d 206 (2007), and Pet. for Cert. in Badgett v. North Carolina, O.T. 2006, No. 07-6156, with Charbonneau, Andre Edwards Sentenced to Life in Prison for 2001 Murder, WRAL, Mar. 26, 2004, online at http://www.wral.com/news/local/story/109648. Why does one defendant who committed a single-victim murder receive the death penalty (due to aggravators of a prior felony conviction and acting recklessly with a gun), while another defendant does not, despite having committed a "triple murder" by killing a young man and his pregnant wife? Compare Commonwealth v. Boxley, 596 Pa. 620 , 948 A.2d 742 (2008), and Pet. for Cert., O.T. 2008, No. 08-6172, with Shea, Judge Gives Consecutive Life Sentences for Triple Murder, Philadelphia Inquirer, June 29, 2004, p. B5. For that matter, why does one defendant who participated in a single-victim murder-for-hire scheme (plus an after-the-fact robbery) receive the death penalty, while another defendant does not, despite having stabbed his wife 60 times and killed his 6-year-old daughter and 3-year-old son while they slept? See Donohue, Capital Punishment in Connecticut, 1973-2007: A Comprehensive Evaluation from 4686 Murders to One Execution, pp. 128-134 (2013), online at http://works.bepress.com/john_ donohue/87. In each instance, the sentences compared were imposed in the same State at about the same time.

The question raised by these examples (and the many more I could give but do not), as well as by the research to which I *2764 have referred, is the same question Justice Stewart, Justice Powell, and others raised over the course of several decades: The imposition and implementation of the death penalty seems capricious, random, indeed, arbitrary. From a defendant's perspective, to receive that sentence, and certainly to find it implemented, is the equivalent of being struck by lightning. How then can we reconcile the death penalty with the demands of a Constitution that first and foremost insists upon a rule of law?

III

"Cruel"-Excessive Delays

The problems of reliability and unfairness almost inevitably lead to a third independent constitutional problem: excessively long periods of time that individuals typically spend on death row, alive but under sentence of death. That is to say, delay is in part a problem that the Constitution's own demands create. Given the special need for reliability and fairness in death penalty cases, the Eighth Amendment does, and must, apply to the death penalty "with special force." Roper, 543 U.S., at 568 , 125 S.Ct. 1183 . Those who face "that most severe sanction must have a fair opportunity to show that the Constitution prohibits their execution." Hall v. Florida, 572 U.S. ----, ----, 134 S.Ct. 1986 , 2001, 188 L.Ed.2d 1007 (2014). At the same time, the Constitution insists that "every safeguard" be "observed" when "a defendant's life is at stake." Gregg, 428 U.S., at 187 , 96 S.Ct. 2909 (joint opinion of Stewart, Powell, and Stevens, JJ.); Furman, 408 U.S., at 306 , 92 S.Ct. 2726 (Stewart, J., concurring) (death "differs from all other forms of criminal punishment, not in degree but in kind"); Woodson, supra, at 305 , 96 S.Ct. 2978 (plurality opinion) ("Death, in its finality, differs more from life imprisonment than a 100-year prison term differs from one of only a year or two").

These procedural necessities take time to implement. And, unless we abandon the procedural requirements that assure fairness and reliability, we are forced to confront the problem of increasingly lengthy delays in capital cases. Ultimately, though these legal causes may help to explain, they do not mitigate the harms caused by delay itself.

A

Consider first the statistics. In 2014, 35 individuals were executed. Those executions occurred, on average, nearly 18 years after a court initially pronounced its sentence of death. DPIC, Execution List 2014, online at http:// www.deathpenaltyinfo.org/execution-list-2014 (showing an average delay of 17 years, 7 months). In some death penalty States, the average delay is longer. In an oral argument last year, for example, the State admitted that the last 10 prisoners executed in Florida had spent an average of nearly 25 years on death row before execution. Tr. of Oral Arg. in Hall v. Florida, O.T. 2013, No. 12-10882, p. 46.

The length of the average delay has increased dramatically over the years. In 1960, the average delay between sentencing and execution was two years. See Aarons, Can Inordinate Delay Between a Death Sentence and Execution Constitute Cruel and Unusual Punishment? 29 Seton Hall L. Rev. 147 , 181 (1998). Ten years ago (in 2004) the average delay was about 11 years. See Dept. of Justice, Bureau of Justice Statistics (BJS), T. Snell, Capital Punishment, 2013-Statistical Tables 14 (Table 10) (rev. Dec. 2014) (hereinafter BJS 2013 Stats). By last year the average had risen to about 18 years. DPIC, Execution List 2014, supra . Nearly half of the 3,000 inmates now on death row have been *2765 there for more than 15 years. And, at present execution rates, it would take more than 75 years to carry out those 3,000 death sentences; thus, the average person on death row would spend an additional 37.5 years there before being executed. BJS 2013 Stats, at 14, 18 (Tables 11 and 15).

I cannot find any reasons to believe the trend will soon be reversed.

B

These lengthy delays create two special constitutional difficulties. See Johnson v. Bredesen, 558 U.S. 1067 , 1069, 130 S.Ct. 541 , 175 L.Ed.2d 552 (2009) (Stevens, J., statement respecting denial of certiorari). First, a lengthy delay in and of itself is especially cruel because it "subjects death row inmates to decades of especially severe, dehumanizing conditions of confinement." Ibid .; Gomez v. Fierro, 519 U.S. 918 , 117 S.Ct. 285 , 136 L.Ed.2d 204 (1996) (Stevens, J., dissenting) (excessive delays from sentencing to execution can themselves "constitute cruel and unusual punishment prohibited by the Eighth Amendment"); see also Lackey v. Texas, 514 U.S. 1045 , 115 S.Ct. 1421 , 131 L.Ed.2d 304 (1995) (memorandum of Stevens, J., respecting denial of certiorari); Knight v. Florida, 528 U.S. 990 , 993, 120 S.Ct. 459 , 145 L.Ed.2d 370 (1999) (BREYER, J., dissenting from denial of certiorari) . Second, lengthy delay undermines the death penalty's penological rationale. Johnson, supra, at 1069, 130 S.Ct. 541 ; Thompson v. McNeil, 556 U.S. 1114 , 1115, 129 S.Ct. 1299 , --- L.Ed.2d ---- (2009) (statement of Stevens, J., respecting denial of certiorari).

Turning to the first constitutional difficulty, nearly all death penalty States keep death row inmates in isolation for 22 or more hours per day. American Civil Liberties Union (ACLU), A Death Before Dying: Solitary Confinement on Death Row 5 (July 2013) (ACLU Report). This occurs even though the ABA has suggested that death row inmates be housed in conditions similar to the general population, and the United Nations Special Rapporteur on Torture has called for a global ban on solitary confinement longer than 15 days. See id., at 2, 4; ABA Standards for Criminal Justice: Treatment of Prisoners 6 (3d ed. 2011). And it is well documented that such prolonged solitary confinement produces numerous deleterious harms. See, e.g., Haney, Mental Health Issues in Long-Term Solitary and "Supermax" Confinement, 49 Crime & Delinquency 124, 130 (2003) (cataloguing studies finding that solitary confinement can cause prisoners to experience "anxiety, panic, rage, loss of control, paranoia, hallucinations, and self-mutilations," among many other symptoms); Grassian, Psychiatric Effects of Solitary Confinement, 22 Wash U. J. L. & Policy 325, 331 (2006) ("[E]ven a few days of solitary confinement will predictably shift the [brain's] electroencephalogram (EEG) pattern toward an abnormal pattern characteristic of stupor and delirium"); accord, In re Medley, 134 U.S. 160 , 167-168, 10 S.Ct. 384 , 33 L.Ed. 835 (1890); see also Davis v. Ayala, --- U.S. ----, 135 S.Ct. 2187 , --- L.Ed.2d ---- (2015) (KENNEDY, J., concurring).

The dehumanizing effect of solitary confinement is aggravated by uncertainty as to whether a death sentence will in fact be carried out. In 1890, this Court recognized that, "when a prisoner sentenced by a court to death is confined in the penitentiary awaiting the execution of the sentence, one of the most horrible feelings to which he can be subjected during that time is the uncertainty during the whole of it." Medley, supra, at 172, 10 S.Ct. 384 . The Court was there describing a delay of a *2766 mere four weeks. In the past century and a quarter, little has changed in this respect-except for duration. Today we must describe delays measured, not in weeks, but in decades. Supra, at 2764 - 2765.

Moreover, we must consider death warrants that have been issued and revoked, not once, but repeatedly. See, e.g., Pet. for Cert. in Suárez Medina v. Texas, O.T. 2001, No. 02-5752, pp. 35-36 (filed Aug. 13, 2002) ("On fourteen separate occasions since Mr. Suárez Medina's death sentence was imposed, he has been informed of the time, date, and manner of his death. At least eleven times, he has been asked to describe the disposal of his bodily remains"); Lithwick, Cruel but not Unusual, Slate, Apr. 1, 2011, online at http://www.slate.com/articles/news_and_politics/jurisprudence/2011/04/cruel_ but_not_unusual.html (John Thompson had seven death warrants signed before he was exonerated); see also, e.g., WFMZ-TV 69 News, Michael John Parrish's Execution Warrant Signed by Governor Corbett (Aug. 18, 2014), online at http:// www.wfmz.com/news/Regional-Poconos-Coal/Local/michael-john-parrishs-execution-warrant-signed-by-governor-corbett/27595356 (former Pennsylvania Governor signed 36 death warrants in his first 3.5 years in office even though Pennsylvania has not carried out an execution since 1999).

Several inmates have come within hours or days of execution before later being exonerated. Willie Manning was four hours from his scheduled execution before the Mississippi Supreme Court stayed the execution. See Robertson, With Hours to Go, Execution is Postponed, N.Y. Times, Apr. 8, 2015, p. A17. Two years later, Manning was exonerated after the evidence against him, including flawed testimony from an FBI hair examiner, was severely undermined. Nave, Why Does the State Still Want to Kill Willie Jerome Manning? Jackson Free Press, Apr. 29, 2015. Nor is Manning an outlier case. See, e.g., Martin, Randall Adams, 61, Dies; Freed With Help of Film, N.Y. Times, June 26, 2011, p. 24 (Randall Adams: stayed by this Court three days before execution; later exonerated); N. Davies, White Lies 231, 292, 298, 399 (1991) (Clarence Lee Brandley: execution stayed twice, once 6 days and once 10 days before; later exonerated); M. Edds, An Expendable Man 93 (2003) (Earl Washington, Jr.: stayed 9 days before execution; later exonerated).

Furthermore, given the negative effects of confinement and uncertainty, it is not surprising that many inmates volunteer to be executed, abandoning further appeals. See, e.g., ACLU Report 8; Rountree, Volunteers for Execution: Directions for Further Research into Grief, Culpability, and Legal Structures, 82 UMKC L. Rev. 295 (2014) (11% of those executed have dropped appeals and volunteered); ACLU Report 3 (account of " 'guys who dropped their appeals because of the intolerable conditions' "). Indeed, one death row inmate, who was later exonerated, still said he would have preferred to die rather than to spend years on death row pursuing his exoneration. Strafer, Volunteering for Execution: Competency, Voluntariness and the Propriety of Third Party Intervention, 74 J. Crim. L. & C. 860, 869 (1983). Nor is it surprising that many inmates consider, or commit, suicide. Id., at 872, n. 44 (35% of those confined on death row in Florida attempted suicide).

Others have written at great length about the constitutional problems that delays create, and, rather than repeat their facts, arguments, and conclusions, I simply refer to some of their writings. See, e.g., Johnson, 558 U.S., at 1069 , 130 S.Ct. 541 (statement of Stevens, J.) (delay "subjects *2767 death row inmates to decades of especially severe, dehumanizing conditions of confinement"); Furman, 408 U.S., at 288 , 92 S.Ct. 2726 (Brennan, J., concurring) ("long wait between the imposition of sentence and the actual infliction of death" is "inevitable" and often "exacts a frightful toll"); Solesbee v. Balkcom, 339 U.S. 9 , 14, 70 S.Ct. 457 , 94 L.Ed. 604 (1950) (Frankfurter, J., dissenting) ("In the history of murder, the onset of insanity while awaiting execution of a death sentence is not a rare phenomenon"); People v. Anderson, 6 Cal.3d 628 , 649, 493 P.2d 880 , 894 (1972) (collecting sources) ("[C]ruelty of capital punishment lies not only in the execution itself and the pain incident thereto, but also in the dehumanizing effects of the lengthy imprisonment prior to execution during which the judicial and administrative procedures essential to due process of law are carried out" (footnote omitted)); District Attorney for Suffolk Dist. v. Watson, 381 Mass. 648 , 673, 411 N.E.2d 1274 , 1287 (1980) (Braucher, J., concurring) (death penalty unconstitutional under State Constitution in part because "[it] will be carried out only after agonizing months and years of uncertainty"); see also Riley v. Attorney General of Jamaica, [1983] 1 A.C. 719, 734-735 (P.C. 1982) (Lord Scarman, joined by Lord Brightman, dissenting) ("execution after inordinate delay" would infringe prohibition against "cruel and unusual punishments" in § 10 of the "Bill of Rights of 1689," the precursor to our Eighth Amendment); Pratt v. Attorney Gen. of Jamaica, [1994] 2 A.C. 1, 4 (P.C. 1993); id., at 32-33 (collecting cases finding inordinate delays unconstitutional or the equivalent); State v . Makwanyane 1995 (3) SA391 (CC) (S. Afr.); Catholic Commission for Justice & Peace in Zimbabwe v. Attorney-General, [1993] 1 Zim. L. R. 242, 282 (inordinate delays unconstitutional); Soering v. United Kingdom, 11 Eur. Ct. H. R. (ser. A), p. 439 (1989) (extradition of murder suspect to United States would violate the European Convention on Human Rights in light of risk of delay before execution); United States v. Burns, [2001] 1 S.C.R. 283 , 353, ¶ 123 (similar).

The second constitutional difficulty resulting from lengthy delays is that those delays undermine the death penalty's penological rationale, perhaps irreparably so. The rationale for capital punishment, as for any punishment, classically rests upon society's need to secure deterrence, incapacitation, retribution, or rehabilitation. Capital punishment by definition does not rehabilitate. It does, of course, incapacitate the offender. But the major alternative to capital punishment-namely, life in prison without possibility of parole-also incapacitates. See Ring v. Arizona, 536 U.S. 584 , 615, 122 S.Ct. 2428 , 153 L.Ed.2d 556 (2002) (BREYER, J., concurring in judgment).

Thus, as the Court has recognized, the death penalty's penological rationale in fact rests almost exclusively upon a belief in its tendency to deter and upon its ability to satisfy a community's interest in retribution. See, e.g., Gregg, 428 U.S., at 183 , 96 S.Ct. 2909 (joint opinion of Stewart, Powell, and Stevens, JJ.). Many studies have examined the death penalty's deterrent effect; some have found such an effect, whereas others have found a lack of evidence that it deters crime. Compare ante, at 2748 - 2749 (SCALIA, J., concurring) (collecting studies finding deterrent effect), with e.g., Sorensen, Wrinkle, Brewer, & Marquart, Capital Punishment and Deterrence: Examining the Effect of Executions on Murder in Texas, 45 Crime & Delinquency 481 (1999) (no evidence of a deterrent effect); Bonner & Fessenden, Absence of Executions: A Special Report, *2768 States With No Death Penalty Share Lower Homicide Rates, N.Y. Times, Sept. 22, 2000, p. A1 (from 1980-2000, homicide rate in death-penalty States was 48% to 101% higher than in non-death-penalty States); Radelet & Akers, Deterrence and the Death Penalty: The Views of the Experts, 87 J. Crim. L. & C. 1, 8 (1996) (over 80% of criminologists believe existing research fails to support deterrence justification); Donohue & Wolfers, Uses and Abuses of Empirical Evidence in the Death Penalty Debate, 58 Stan. L. Rev. 791 , 794 (2005) (evaluating existing statistical evidence and concluding that there is "profound uncertainty" about the existence of a deterrent effect).

Recently, the National Research Council (whose members are drawn from the councils of the National Academy of Sciences, the National Academy of Engineering, and the Institute of Medicine) reviewed 30 years of empirical evidence and concluded that it was insufficient to establish a deterrent effect and thus should "not be used to inform" discussion about the deterrent value of the death penalty. National Research Council, Deterrence and the Death Penalty 2 (D. Nagin & J. Pepper eds. 2012); accord, Baze v. Rees, 553 U.S. 35 , 79, 128 S.Ct. 1520 , 170 L.Ed.2d 420 (2008) (Stevens, J., concurring in judgment) ("Despite 30 years of empirical research in the area, there remains no reliable statistical evidence that capital punishment in fact deters potential offenders").

I recognize that a "lack of evidence" for a proposition does not prove the contrary. See Ring, supra, at 615 , 122 S.Ct. 2428 (one might believe the studies "inconclusive"). But suppose that we add to these studies the fact that, today, very few of those sentenced to death are actually executed, and that even those executions occur, on average, after nearly two decades on death row. DPIC, Execution List 2014, supra . Then, does it still seem likely that the death penalty has a significant deterrent effect?

Consider, for example, what actually happened to the 183 inmates sentenced to death in 1978. As of 2013 (35 years later), 38 (or 21% of them) had been executed; 132 (or 72%) had had their convictions or sentences overturned or commuted; and 7 (or 4%) had died of other (likely natural) causes. Six (or 3%) remained on death row. BJS 2013 Stats, at 19 (Table 16).

The example illustrates a general trend. Of the 8,466 inmates under a death sentence at some point between 1973 and 2013, 16% were executed, 42% had their convictions or sentences overturned or commuted, and 6% died by other causes; the remainder (35%) are still on death row. Id., at 20 (Table 17); see also Baumgartner & Dietrich, Most Death Penalty Sentences Are Overturned: Here's Why That Matters, Washington Post Blog, Monkey Cage, Mar. 17, 2015 (similar).

Thus an offender who is sentenced to death is two or three times more likely to find his sentence overturned or commuted than to be executed; and he has a good chance of dying from natural causes before any execution (or exoneration) can take place. In a word, executions are rare . And an individual contemplating a crime but evaluating the potential punishment would know that, in any event, he faces a potential sentence of life without parole.

These facts, when recurring, must have some offsetting effect on a potential perpetrator's fear of a death penalty. And, even if that effect is no more than slight, it makes it difficult to believe (given the studies of deterrence cited earlier) that such a rare event significantly deters horrendous crimes. See Furman, 408 U.S., at 311-312 , 92 S.Ct. 2726 (White, J., concurring) (It cannot "be said with confidence that society's need for specific deterrence *2769 justifies death for so few when for so many in like circumstances life imprisonment or shorter prison terms are judged sufficient").

But what about retribution? Retribution is a valid penological goal. I recognize that surviving relatives of victims of a horrendous crime, or perhaps the community itself, may find vindication in an execution. And a community that favors the death penalty has an understandable interest in representing their voices. But see A. Sarat, Mercy on Trial: What It Means To Stop an Execution 130 (2005) (Illinois Governor George Ryan explained his decision to commute all death sentences on the ground that it was "cruel and unusual" for "family members to go through this ... legal limbo for [20] years").

The relevant question here, however, is whether a "community's sense of retribution" can often find vindication in "a death that comes," if at all, "only several decades after the crime was committed." Valle v. Florida, 564 U.S. ----, ----, 132 S.Ct. 1 , 2, 180 L.Ed.2d 940 (2011) (BREYER, J., dissenting from denial of stay). By then the community is a different group of people. The offenders and the victims' families have grown far older. Feelings of outrage may have subsided. The offender may have found himself a changed human being. And sometimes repentance and even forgiveness can restore meaning to lives once ruined. At the same time, the community and victims' families will know that, even without a further death, the offender will serve decades in prison under a sentence of life without parole.

I recognize, of course, that this may not always be the case, and that sometimes the community believes that an execution could provide closure. Nevertheless, the delays and low probability of execution must play some role in any calculation that leads a community to insist on death as retribution. As I have already suggested, they may well attenuate the community's interest in retribution to the point where it cannot by itself amount to a significant justification for the death penalty. Id., at ----, 132 S.Ct., at 2 . In any event, I believe that whatever interest in retribution might be served by the death penalty as currently administered, that interest can be served almost as well by a sentence of life in prison without parole (a sentence that every State now permits, see ACLU, A Living Death: Life Without Parole for Nonviolent Offenses 11, and n. 10 (2013)).

Finally, the fact of lengthy delays undermines any effort to justify the death penalty in terms of its prevalence when the Founders wrote the Eighth Amendment. When the Founders wrote the Constitution, there were no 20- or 30-year delays. Execution took place soon after sentencing. See P. Mackey, Hanging in the Balance: The Anti-Capital Punishment Movement in New York State, 1776-1861, p. 17 (1982); T. Jefferson, A Bill for Proportioning Crimes and Punishments (1779), reprinted in The Complete Jefferson 90, 95 (S. Padover ed. 1943); 2 Papers of John Marshall 207-209 (C. Cullen & H. Johnson eds. 1977) (describing petition for commutation based in part on 5-month delay); Pratt v. Attorney Gen. of Jamaica, [1994] 2 A. C., at 17 (same in United Kingdom) (collecting cases). And, for reasons I shall describe, infra, at 2770 - 2773, we cannot return to the quick executions in the founding era.

The upshot is that lengthy delays both aggravate the cruelty of the death penalty and undermine its jurisprudential rationale. And this Court has said that, if the death penalty does not fulfill the goals of deterrence or retribution, "it is nothing more than the purposeless and needless imposition of pain and suffering and hence *2770 an unconstitutional punishment." Atkins, 536 U.S., at 319 , 122 S.Ct. 2242 (quoting Enmund v. Florida, 458 U.S. 782 , 798, 102 S.Ct. 3368 , 73 L.Ed.2d 1140 (1982); internal quotation marks omitted); see also Gregg, 428 U.S., at 183 , 96 S.Ct. 2909 (joint opinion of Stewart, Powell, and Stevens, JJ.) ("sanction imposed cannot be so totally without penological justification that it results in the gratuitous infliction of suffering"); Furman, supra, at 312 , 92 S.Ct. 2726 (White, J., concurring) (a "penalty with such negligible returns to the State would be patently excessive and cruel and unusual punishment violative of the Eighth Amendment"); Thompson, 556 U.S., at 1115 , 129 S.Ct. 1299 (statement of Stevens, J., respecting denial of certiorari) (similar).

Indeed, Justice Lewis Powell (who provided a crucial vote in Gregg ) came to much the same conclusion, albeit after his retirement from this Court. Justice Powell had come to the Court convinced that the Federal Constitution did not outlaw the death penalty but rather left the matter up to individual States to determine. Furman, supra, at 431-432, 92 S.Ct. 2726 (Powell, J., dissenting); see also J. Jeffries, Justice Lewis F. Powell, Jr., p. 409 (2001) (describing Powell, during his time on the Court, as a "fervent partisan" of "the constitutionality of capital punishment").

Soon after Justice Powell's retirement, Chief Justice Rehnquist appointed him to chair a committee addressing concerns about delays in capital cases, the Ad Hoc Committee on Federal Habeas Corpus in Capital Cases (Committee). The Committee presented a report to Congress, and Justice Powell testified that "[d]elay robs the penalty of much of its deterrent value." Habeas Corpus Reform, Hearings before the Senate Committee on the Judiciary, 100th Cong., 1st and 2d Sess., 35 (1989 and 1990). Justice Powell, according to his official biographer, ultimately concluded that capital punishment:

" 'serves no useful purpose.' The United States was 'unique among the industrialized nations of the West in maintaining the death penalty,' and it was enforced so rarely that it could not deter. More important, the haggling and delay and seemingly endless litigation in every capital case brought the law itself into disrepute." Jeffries, supra, at 452.

In short, the problem of excessive delays led Justice Powell, at least in part, to conclude that the death penalty was unconstitutional.

As I have said, today delays are much worse. When Chief Justice Rehnquist appointed Justice Powell to the Committee, the average delay between sentencing and execution was 7 years and 11 months, compared with 17 years and 7 months today. Compare BJS, L. Greenfeld, Capital Punishment, 1990, p. 11 (Table 12) (Sept. 1991) with supra, at 18-19.

C

One might ask, why can Congress or the States not deal directly with the delay problem? Why can they not take steps to shorten the time between sentence and execution, and thereby mitigate the problems just raised? The answer is that shortening delay is much more difficult than one might think. And that is in part because efforts to do so risk causing procedural harms that also undermine the death penalty's constitutionality.

For one thing, delays have helped to make application of the death penalty more reliable. Recall the case of Henry Lee McCollum, whom DNA evidence exonerated 30 years after his conviction. Katz & Eckholm, N.Y. Times, at A1. If McCollum had been executed earlier, he would not have lived to see the day when DNA evidence exonerated him and implicated *2771 another man; that man is already serving a life sentence for a rape and murder that he committed just a few weeks after the murder McCollum was convicted of. Ibid. In fact, this Court had earlier denied review of McCollum's claim over the public dissent of only one Justice. McCollum v. North Carolina, 512 U.S. 1254 , 114 S.Ct. 2784 , 129 L.Ed.2d 895 (1994). And yet a full 20 years after the Court denied review, McCollum was exonerated by DNA evidence. There are a significant number of similar cases, some of which I have discussed earlier. See also DPIC Innocence List, supra (Nathson Fields, 23 years; Paul House, 23 years; Nicholas Yarris, 21 years; Anthony Graves, 16 years; Damon Thibodeaux, 15 years; Ricky Jackson, Wiley Bridgeman, and Kwame Ajamu, all exonerated for the same crime 39 years after their convictions).

In addition to those who are exonerated on the ground that they are innocent, there are other individuals whose sentences or convictions have been overturned for other reasons (as discussed above, state and federal courts found error in 68% of the capital cases they reviewed between 1973 and 1995). See Part I, supra . In many of these cases, a court will have found that the individual did not merit the death penalty in a special sense-namely, he failed to receive all the procedural protections that the law requires for the death penalty's application. By eliminating some of these protections, one likely could reduce delay. But which protections should we eliminate? Should we eliminate the trial-related protections we have established for capital defendants: that they be able to present to the sentencing judge or jury all mitigating circumstances, Lockett v. Ohio, 438 U.S. 586 , 98 S.Ct. 2954 , 57 L.Ed.2d 973 ; that the State provide guidance adequate to reserve the application of the death penalty to particularly serious murders, Gregg, 428 U.S. 153 , 96 S.Ct. 2909 , 49 L.Ed.2d 859 ; that the State provide adequate counsel and, where warranted, adequate expert assistance, Powell v. Alabama, 287 U.S. 45 , 53 S.Ct. 55 , 77 L.Ed. 158 (1932); Wiggins v. Smith, 539 U.S. 510 , 123 S.Ct. 2527 , 156 L.Ed.2d 471 (2003); Ake v. Oklahoma, 470 U.S. 68 , 105 S.Ct. 1087 , 84 L.Ed.2d 53 (1985); or that a jury must find the aggravating factors necessary to impose the death penalty, Ring, 536 U.S. 584 , 122 S.Ct. 2428 , 153 L.Ed.2d 556 ; see also id., at 614 , 122 S.Ct. 2428 (BREYER, J., concurring in judgment)? Should we no longer ensure that the State does not execute those who are seriously intellectually disabled, Atkins, 536 U.S. 304 , 122 S.Ct. 2242 , 153 L.Ed.2d 335 ? Should we eliminate the requirement that the manner of execution be constitutional, Baze, 553 U.S. 35 , 128 S.Ct. 1520 , 170 L.Ed.2d 420 , or the requirement that the inmate be mentally competent at the time of his execution, Ford v. Wainwright, 477 U.S. 399 , 106 S.Ct. 2595 , 91 L.Ed.2d 335 (1986)? Or should we get rid of the criminal protections that all criminal defendants receive-for instance, that defendants claiming violation of constitutional guarantees (say "due process of law") may seek a writ of habeas corpus in federal courts? See, e.g., O'Neal v. McAninch, 513 U.S. 432 , 115 S.Ct. 992 , 130 L.Ed.2d 947 (1995). My answer to these questions is "surely not." But see ante, at 2748 - 2750 (SCALIA, J., concurring).

One might, of course, argue that courts, particularly federal courts providing additional layers of review, apply these and other requirements too strictly, and that causes delay. But, it is difficult for judges, as it would be difficult for anyone, not to apply legal requirements punctiliously when the consequence of failing to do so may well be death, particularly the death of an innocent person. See, e.g., *2772 Zant v. Stephens, 462 U.S. 862 , 885, 103 S.Ct. 2733 , 77 L.Ed.2d 235 (1983) ("[A]lthough not every imperfection in the deliberative process is sufficient, even in a capital case, to set aside a state-court judgment, the severity of the sentence mandates careful scrutiny in the review of any colorable claim of error"); Kyles v. Whitley, 514 U.S. 419 , 422, 115 S.Ct. 1555 , 131 L.Ed.2d 490 (1995) ("[O]ur duty to search for constitutional error with painstaking care is never more exacting than it is in a capital case" (internal quotation marks omitted)); Thompson, 556 U.S., at 1116 , 129 S.Ct. 1299 (statement of Stevens, J.) ("Judicial process takes time, but the error rate in capital cases illustrates its necessity").

Moreover, review by courts at every level helps to ensure reliability; if this Court had not ordered that Anthony Ray Hinton receive further hearings in state court, see Hinton v. Alabama, 571 U.S. ----, 134 S.Ct. 1081 , 188 L.Ed.2d 1 , he may well have been executed rather than exonerated. In my own view, our legal system's complexity, our federal system with its separate state and federal courts, our constitutional guarantees, our commitment to fair procedure, and, above all, a special need for reliability and fairness in capital cases, combine to make significant procedural "reform" unlikely in practice to reduce delays to an acceptable level.

And that fact creates a dilemma: A death penalty system that seeks procedural fairness and reliability brings with it delays that severely aggravate the cruelty of capital punishment and significantly undermine the rationale for imposing a sentence of death in the first place. See Knight, 528 U.S., at 998 , 120 S.Ct. 459 (BREYER, J., dissenting from denial of certiorari) (one of the primary causes of the delay is the States' "failure to apply constitutionally sufficient procedures at the time of initial [conviction or] sentencing"). But a death penalty system that minimizes delays would undermine the legal system's efforts to secure reliability and procedural fairness.

In this world, or at least in this Nation, we can have a death penalty that at least arguably serves legitimate penological purposes or we can have a procedural system that at least arguably seeks reliability and fairness in the death penalty's application. We cannot have both. And that simple fact, demonstrated convincingly over the past 40 years, strongly supports the claim that the death penalty violates the Eighth Amendment. A death penalty system that is unreliable or procedurally unfair would violate the Eighth Amendment. Woodson, 428 U.S., at 305, 96 S.Ct. 2978 (plurality opinion); Hall, 572 U.S., at ----, 134 S.Ct., at 2001; Roper, 543 U.S., at 568 , 125 S.Ct. 1183 . And so would a system that, if reliable and fair in its application of the death penalty, would serve no legitimate penological purpose. Furman, 408 U.S., at 312 , 92 S.Ct. 2726 (White, J., concurring); Gregg, supra, at 183 , 96 S.Ct. 2909 (joint opinion of Stewart, Powell, and Stevens, JJ.); Atkins, supra, at 319, 122 S.Ct. 2242 .

IV

"Unusual"-Decline in Use of the Death Penalty

The Eighth Amendment forbids punishments that are cruel and unusual. Last year, in 2014, only seven States carried out an execution. Perhaps more importantly, in the last two decades, the imposition and implementation of the death penalty have increasingly become unusual. I can illustrate the significant decline in the use of the death penalty in several ways.

An appropriate starting point concerns the trajectory of the number of annual death sentences nationwide, from the 1970's to present day. In 1977-just after *2773 the Supreme Court made clear that, by modifying their legislation, States could reinstate the death penalty-137 people were sentenced to death. BJS 2013 Stats, at 19 (Table 16). Many States having revised their death penalty laws to meet Furman' s requirements, the number of death sentences then increased. Between 1986 and 1999, 286 persons on average were sentenced to death each year. BJS 2013 Stats, at 14, 19 (Tables 11 and 16). But, approximately 15 years ago, the numbers began to decline, and they have declined rapidly ever since. See Appendix A, infra (showing sentences from 1977-2014). In 1999, 279 persons were sentenced to death. BJS 2013 Stats, at 19 (Table 16). Last year, just 73 persons were sentenced to death. DPIC, The Death Penalty in 2014: Year End Report 1 (2015).

That trend, a significant decline in the last 15 years, also holds true with respect to the number of annual executions. See Appendix B, infra (showing executions from 1977-2014). In 1999, 98 people were executed. BJS, Data Collection: National Prisoner Statistics Program (BJS Prisoner Statistics) (available in Clerk of Court's case file). Last year, that number was only 35. DPIC, The Death Penalty in 2014, supra, at 1.

Next, one can consider state-level data. Often when deciding whether a punishment practice is, constitutionally speaking, "unusual," this Court has looked to the number of States engaging in that practice. Atkins, 536 U.S., at 313-316 , 122 S.Ct. 2242 ; Roper, supra, at 564-566, 125 S.Ct. 1183 . In this respect, the number of active death penalty States has fallen dramatically. In 1972, when the Court decided Furman , the death penalty was lawful in 41 States. Nine States had abolished it. E. Mandery, A Wild Justice: The Death and Resurrection of Capital Punishment in America 145 (2013). As of today, 19 States have abolished the death penalty (along with the District of Columbia), although some did so prospectively only. See DPIC, States With and Without the Death Penalty, online at http://www.deathpenalty info.org/states-and-without-death-penalty. In 11 other States that maintain the death penalty on the books, no execution has taken place for more than eight years: Arkansas (last execution 2005); California (2006); Colorado (1997); Kansas (no executions since the death penalty was reinstated in 1976); Montana (2006); Nevada (2006); New Hampshire (no executions since the death penalty was reinstated in 1976); North Carolina (2006); Oregon (1997); Pennsylvania (1999); and Wyoming (1992). DPIC, Executions by State and Year, online at http://www.death penaltyinfo.org/node/5741.

Accordingly, 30 States have either formally abolished the death penalty or have not conducted an execution in more than eight years. Of the 20 States that have conducted at least one execution in the past eight years, 9 have conducted fewer than five in that time, making an execution in those States a fairly rare event. BJS Prisoner Statistics (Delaware, Idaho, Indiana, Kentucky, Louisiana, South Dakota, Tennessee, Utah, Washington). That leaves 11 States in which it is fair to say that capital punishment is not "unusual." And just three of those States (Texas, Missouri, and Florida) accounted for 80% of the executions nationwide (28 of the 35) in 2014. See DPIC, Number of Executions by State and Region Since 1976, online at http://www.deathpenalty info.org/number-executions-state-and-region-1976. Indeed, last year, only seven States conducted an execution. DPIC, Executions by State and Year, supra ; DPIC, Death Sentences in the United States From 1977 by State and by Year, online at http://www.deathpenaltyinfo.org/death-sentences-united-states-1977-2008. In *2774 other words, in 43 States, no one was executed.

In terms of population, if we ask how many Americans live in a State that at least occasionally carries out an execution (at least one within the prior three years), the answer two decades ago was 60% or 70%. Today, that number is 33%. See Appendix C, infra .

At the same time, use of the death penalty has become increasingly concentrated geographically. County-by-county figures are relevant, for decisions to impose the death penalty typically take place at a county level. See supra, at 2761 - 2762. County-level sentencing figures show that, between 1973 and 1997, 66 of America's 3,143 counties accounted for approximately 50% of all death sentences imposed. Liebman & Clarke 264-265; cf. id., at 266 . (counties with 10% of the Nation's population imposed 43% of its death sentences). By the early 2000's, the death penalty was only actively practiced in a very small number of counties: between 2004 and 2009, only 35 counties imposed 5 or more death sentences, i.e., approximately one per year. See Appendix D, infra (such counties colored in red) (citing Ford, The Death Penalty's Last Stand, The Atlantic, Apr. 21, 2015). And more recent data show that the practice has diminished yet further: between 2010 and 2015 (as of June 22), only 15 counties imposed five or more death sentences. See Appendix E, infra . In short, the number of active death penalty counties is small and getting smaller. And the overall statistics on county-level executions bear this out. Between 1976 and 2007, there were no executions in 86% of America's counties. Liebman & Clarke 265-266, and n. 47; cf. ibid. (counties with less than 5% of the Nation's population carried out over half of its executions from 1976-2007).

In sum, if we look to States, in more than 60% there is effectively no death penalty, in an additional 18% an execution is rare and unusual, and 6%, i.e., three States, account for 80% of all executions. If we look to population, about 66% of the Nation lives in a State that has not carried out an execution in the last three years. And if we look to counties, in 86% there is effectively no death penalty. It seems fair to say that it is now unusual to find capital punishment in the United States, at least when we consider the Nation as a whole. See Furman, 408 U.S., at 311 , 92 S.Ct. 2726 (1972) (White, J., concurring) (executions could be so infrequently carried out that they "would cease to be a credible deterrent or measurably to contribute to any other end of punishment in the criminal justice system ... when imposition of the penalty reaches a certain degree of infrequency, it would be very doubtful that any existing general need for retribution would be measurably satisfied").

Moreover, we have said that it " 'is not so much the number of these States that is significant, but the consistency of the direction of change.' " Roper, 543 U.S., at 566 , 125 S.Ct. 1183 (quoting Atkins, supra, at 315, 122 S.Ct. 2242 ) (finding significant that five States had abandoned the death penalty for juveniles, four legislatively and one judicially, since the Court's decision in Stanford v. Kentucky, 492 U.S. 361 , 109 S.Ct. 2969 , 106 L.Ed.2d 306 (1989)). Judged in that way, capital punishment has indeed become unusual. Seven States have abolished the death penalty in the last decade, including (quite recently) Nebraska. DPIC, States With and Without the Death Penalty, supra . And several States have come within a single vote of eliminating the death penalty. Seelye, Measure to Repeal Death Penalty Fails by a Single Vote in New Hampshire Senate, N.Y. Times, Apr. 17, 2014, p. A12; Dennison, House Deadlocks on Bill To Abolish *2775 Death Penalty in Montana, Billings Gazette, Feb. 23, 2015; see also Offredo, Delaware Senate Passes Death Penalty Repeal Bill, Delaware News Journal, Apr. 3, 2015. Eleven States, as noted earlier, have not executed anyone in eight years. Supra, at 2773 - 2774. And several States have formally stopped executing inmates. See Yardley, Oregon's Governor Says He Will Not Allow Executions, N.Y. Times, Nov. 23, 2011, p. A14 (Oregon); Governor of Colorado, Exec. Order No. D2013-006, May 22, 2013 (Colorado); Lovett, Executions Are Suspended by Governor in Washington, N.Y. Times, Feb. 12, 2014, p. A12 (Washington); Begley, Pennsylvania Stops Using the Death Penalty, Time, Feb. 13, 2015 (Pennsylvania); see also Welsh-Huggins, Associated Press, Ohio Executions Rescheduled, Jan. 30, 2015 (Ohio).

Moreover, the direction of change is consistent. In the past two decades, no State without a death penalty has passed legislation to reinstate the penalty. See Atkins, supra, at 315-316, 122 S.Ct. 2242 ; DPIC, States With and Without the Death Penalty, supra . Indeed, even in many States most associated with the death penalty, remarkable shifts have occurred. In Texas, the State that carries out the most executions, the number of executions fell from 40 in 2000 to 10 in 2014, and the number of death sentences fell from 48 in 1999 to 9 in 2013 (and 0 thus far in 2015). DPIC, Executions by State and Year, supra ; BJS, T. Snell, Capital Punishment, 1999, p. 6 (Table 5) (Dec. 2000) (hereinafter BJS 1999 Stats); BJS 2013 Stats, at 19 (Table 16); von Drehle, Bungled Executions, Backlogged Courts, and Three More Reasons the Modern Death Penalty Is a Failed Experiment, Time, June 8, 2015, p. 26. Similarly dramatic declines are present in Virginia, Oklahoma, Missouri, and North Carolina. BJS 1999 Stats, at 6 (Table 5); BJS 2013 Stats, at 19 (Table 16).

These circumstances perhaps reflect the fact that a majority of Americans, when asked to choose between the death penalty and life in prison without parole, now choose the latter. Wilson, Support for Death Penalty Still High, But Down, Washington Post, GovBeat, June 5, 2014, online at www. washingtonpost.com/blogs/govbeat/wp/2014/06/05/support-for-death-penalty-still-high-but-down; see also ALI, Report of the Council to the Membership on the Matter of the Death Penalty 4 (Apr. 15, 2009) (withdrawing Model Penal Code section on capital punishment section from the Code, in part because of doubts that the American Law Institute could "recommend procedures that would" address concerns about the administration of the death penalty); cf. Gregg, 428 U.S., at 193-194, 96 S.Ct. 2909 (joint opinion of Stewart, Powell, and Stevens, JJ.) (relying in part on Model Penal Code to conclude that a "carefully drafted statute" can satisfy the arbitrariness concerns expressed in Furman ).

I rely primarily upon domestic, not foreign events, in pointing to changes and circumstances that tend to justify the claim that the death penalty, constitutionally speaking, is "unusual." Those circumstances are sufficient to warrant our reconsideration of the death penalty's constitutionality. I note, however, that many nations-indeed, 95 of the 193 members of the United Nations-have formally abolished the death penalty and an additional 42 have abolished it in practice. Oakford, UN Vote Against Death Penalty Highlights Global Abolitionist Trend-and Leaves the U.S. Stranded, Vice News, Dec. 19, 2014, online at https://news.vice.com/article/un-vote-against-death-penalty-highlights-global-abolitionist-trend-and-leaves-the-us-stranded. In 2013, only 22 countries in the world carried out an execution. International Commission Against *2776 Death Penalty, Review 2013, pp. 2-3. No executions were carried out in Europe or Central Asia, and the United States was the only country in the Americas to execute an inmate in 2013. Id., at 3. Only eight countries executed more than 10 individuals (the United States, China, Iran, Iraq, Saudi Arabia, Somalia, Sudan, Yemen). Id., at 2. And almost 80% of all known executions took place in three countries: Iran, Iraq, and Saudi Arabia. Amnesty International, Death Sentences and Executions 2013, p. 3 (2014). (This figure does not include China, which has a large population, but where precise data cannot be obtained. Id., at 2.)

V

I recognize a strong counterargument that favors constitutionality. We are a court. Why should we not leave the matter up to the people acting democratically through legislatures? The Constitution foresees a country that will make most important decisions democratically. Most nations that have abandoned the death penalty have done so through legislation, not judicial decision. And legislators, unlike judges, are free to take account of matters such as monetary costs, which I do not claim are relevant here. See, e.g., Berman, Nebraska Lawmakers Abolish the Death Penalty, Narrowly Overriding Governor's Veto, Washington Post Blog, Post Nation, May 27, 2015 (listing cost as one of the reasons why Nebraska legislators recently repealed the death penalty in that State); cf. California Commission on the Fair Administration of Justice, Report and Recommendations on the Administration of the Death Penalty in California 117 (June 30, 2008) (death penalty costs California $137 million per year; a comparable system of life imprisonment without parole would cost $11.5 million per year), online at http://www.ccfaj.org/rr-dp-official.html; Dáte, The High Price of Killing Killers, Palm Beach Post, Jan. 4, 2000, p. 1A (cost of each execution is $23 million above cost of life imprisonment without parole in Florida).

The answer is that the matters I have discussed, such as lack of reliability, the arbitrary application of a serious and irreversible punishment, individual suffering caused by long delays, and lack of penological purpose are quintessentially judicial matters. They concern the infliction-indeed the unfair, cruel, and unusual infliction-of a serious punishment upon an individual. I recognize that in 1972 this Court, in a sense, turned to Congress and the state legislatures in its search for standards that would increase the fairness and reliability of imposing a death penalty. The legislatures responded. But, in the last four decades, considerable evidence has accumulated that those responses have not worked.

Thus we are left with a judicial responsibility. The Eighth Amendment sets forth the relevant law, and we must interpret that law. See Marbury v. Madison, 1 Cranch 137 , 177, 2 L.Ed. 60 (1803); Hall, 572 U.S., at ----, 134 S.Ct., at 2000 ("That exercise of independent judgment is the Court's judicial duty"). We have made clear that " 'the Constitution contemplates that in the end our own judgment will be brought to bear on the question of the acceptability of the death penalty under the Eighth Amendment.' " Id., at ----, 134 S.Ct., at 1999 (quoting Coker v. Georgia, 433 U.S. 584 , 597, 97 S.Ct. 2861 , 53 L.Ed.2d 982 (1977) (plurality opinion)); see also Thompson v. Oklahoma, 487 U.S. 815 , 833, n. 40, 108 S.Ct. 2687 , 101 L.Ed.2d 702 (1988) (plurality opinion).

For the reasons I have set forth in this opinion, I believe it highly likely that the *2777 death penalty violates the Eighth Amendment. At the very least, the Court should call for full briefing on the basic question.

With respect, I dissent.

Appendix A

APPENDICES

A

Death Sentences Imposed 1977-2014

?

Appendix B

B

Executions 1977-2014

?

*2778 Appendix C

C

Percentage of U.S. population in States that conducted an execution within prior 3 years

?

*2779 Appendix D

?

*2780 Appendix E

?

Justice SOTOMAYOR, with whom Justice GINSBURG, Justice BREYER, and Justice KAGAN join, dissenting.

Petitioners, three inmates on Oklahoma's death row, challenge the constitutionality of the State's lethal injection protocol. The State plans to execute petitioners using three drugs: midazolam, rocuronium bromide, and potassium chloride. The latter two drugs are intended to paralyze the inmate and stop his heart.

*2781 But they do so in a torturous manner, causing burning, searing pain. It is thus critical that the first drug, midazolam, do what it is supposed to do, which is to render and keep the inmate unconscious. Petitioners claim that midazolam cannot be expected to perform that function, and they have presented ample evidence showing that the State's planned use of this drug poses substantial, constitutionally intolerable risks.

Nevertheless, the Court today turns aside petitioners' plea that they at least be allowed a stay of execution while they seek to prove midazolam's inadequacy. The Court achieves this result in two ways: first, by deferring to the District Court's decision to credit the scientifically unsupported and implausible testimony of a single expert witness; and second, by faulting petitioners for failing to satisfy the wholly novel requirement of proving the availability of an alternative means for their own executions. On both counts the Court errs. As a result, it leaves petitioners exposed to what may well be the chemical equivalent of being burned at the stake.

I

A

The Eighth Amendment succinctly prohibits the infliction of "cruel and unusual punishments." Seven years ago, in Baze v. Rees, 553 U.S. 35 , 128 S.Ct. 1520 , 170 L.Ed.2d 420 (2008), the Court addressed the application of this mandate to Kentucky's lethal injection protocol. At that time, Kentucky, like at least 29 of the 35 other States with the death penalty, utilized a series of three drugs to perform executions: (1) sodium thiopental, a "fast-acting barbiturate sedative that induces a deep, comalike unconsciousness when given in the amounts used for lethal injection"; (2) pancuronium bromide, "a paralytic agent that inhibits all muscular-skeletal movements and ... stops respiration"; and (3) potassium chloride, which "interferes with the electrical signals that stimulate the contractions of the heart, inducing cardiac arrest." Id., at 44 , 128 S.Ct. 1520 (plurality opinion of ROBERTS, C.J.).

In Baze, it was undisputed that absent a "proper dose of sodium thiopental," there would be a "substantial, constitutionally unacceptable risk of suffocation from the administration of pancuronium bromide and pain from the injection of potassium chloride." Id., at 53 , 128 S.Ct. 1520 . That is because, if given to a conscious inmate, pancuronium bromide would leave him or her asphyxiated and unable to demonstrate "any outward sign of distress," while potassium chloride would cause "excruciating pain." Id., at 71 , 128 S.Ct. 1520 (Stevens, J., concurring in judgment). But the Baze petitioners conceded that if administered as intended, Kentucky's method of execution would nevertheless "result in a humane death," id., at 41 , 128 S.Ct. 1520 (plurality opinion), as the "proper administration" of sodium thiopental "eliminates any meaningful risk that a prisoner would experience pain from the subsequent injections of pancuronium and potassium chloride," id., at 49 , 128 S.Ct. 1520 . Based on that premise, the Court ultimately rejected the challenge to Kentucky's protocol, with the plurality opinion concluding that the State's procedures for administering these three drugs ensured there was no "objectively intolerable risk" of severe pain. Id., at 61-62 , 128 S.Ct. 1520 (internal quotation marks omitted).

B

For many years, Oklahoma performed executions using the same three drugs at issue in Baze . After Baze was decided, however, the primary producer of sodium thiopental refused to continue permitting *2782 the drug to be used in executions. Ante, at 2756 - 2757. Like a number of other States, Oklahoma opted to substitute pentobarbital, another barbiturate, in its place. But in March 2014, shortly before two scheduled executions, Oklahoma found itself unable to secure this drug. App. 144.

The State rescheduled the executions for the following month to give it time to locate an alternative anesthetic. In less than a week, a group of officials from the Oklahoma Department of Corrections and the Attorney General's office selected midazolam to serve as a replacement for pentobarbital. Id., at 145, 148-149 .

Soon thereafter, Oklahoma used midazolam for the first time in its execution of Clayton Lockett. That execution did not go smoothly. Ten minutes after an intravenous (IV) line was set in Lockett's groin area and 100 milligrams of midazolam were administered, an attending physician declared Lockett unconscious. Id., at 392-393. When the paralytic and potassium chloride were administered, however, Lockett awoke. Ibid . Various witnesses reported that Lockett began to writhe against his restraints, saying, "[t]his s* * * is f* * *ing with my mind," "something is wrong," and "[t]he drugs aren't working." Id., at 53 (internal quotation marks omitted). State officials ordered the blinds lowered, then halted the execution. Id., at 393, 395. But 10 minutes later-approximately 40 minutes after the execution began-Lockett was pronounced dead. Id., at 395.

The State stayed all future executions while it sought to determine what had gone wrong in Lockett's. Five months later, the State released an investigative report identifying a flaw in the IV line as the principal difficulty: The IV had failed to fully deliver the lethal drugs into Lockett's veins. Id., at 398. An autopsy determined, however, that the concentration of midazolam in Lockett's blood was more than sufficient to render an average person unconscious. Id., at 397, 405.

In response to this report, the State modified its lethal injection protocol. The new protocol contains a number of procedures designed to guarantee that members of the execution team are able to insert the IV properly, and charges them with ensuring that the inmate is unconscious. Id., at 57-66, 361-369. But the protocol continues to authorize the use of the same three-drug formula used to kill Lockett-though it does increase the intended dose of midazolam from 100 milligrams to 500 milligrams. Id., at 61. The State has indicated that it plans to use this drug combination in all upcoming executions, subject to only an immaterial substitution of paralytic agents. Ante, at 2758 - 2759.

C

In June 2014, inmates on Oklahoma's death row filed a 42 U.S.C. § 1983 suit against respondent prison officials challenging the constitutionality of Oklahoma's method of execution. After the State released its revised execution protocol, the four inmates whose executions were most imminent-Charles Warner, along with petitioners Richard Glossip, John Grant, and Benjamin Cole-moved for a preliminary injunction. They contended, among other things, that the State's intended use of midazolam would violate the Eighth Amendment because, unlike sodium thiopental or pentobarbital, the drug "is incapable of producing a state of unawareness that will be reliably maintained after either of the other two pain-producing drugs ... is injected." Amended Complaint ¶ 101.

The District Court held a 3-day evidentiary hearing, at which petitioners relied principally on the testimony of two experts: Dr. David Lubarsky, an anesthesiologist, *2783 and Dr. Larry Sasich, a doctor of pharmacy. The State, in turn, based its case on the testimony of Dr. Roswell Evans, also a doctor of pharmacy.

To a great extent, the experts' testimony overlapped. All three experts agreed that midazolam is from a class of sedative drugs known as benzodiazepines (a class that includes Valium and Xanax), and that it has no analgesic-or pain-relieving-effects. App. 205 (Lubarsky), 260-261 (Sasich), 311 (Evans). They further agreed that while midazolam can be used to render someone unconscious, it is not approved by the Federal Drug Administration (FDA) for use as, and is not in fact used as, a "sole drug to produce and maintain anesthesia in surgical proceedings." Id., at 307, 327 (Evans); see id., at 171 (Lubarsky); id., at 262 (Sasich). Finally, all three experts recognized that midazolam is subject to a ceiling effect, which means that there is a point at which increasing the dose of the drug does not result in any greater effect. Id., at 172 (Lubarsky), 243 (Sasich), 331 (Evans).

The experts' opinions diverged, however, on the crucial questions of how this ceiling effect operates, and whether it will prevent midazolam from keeping a condemned inmate unconscious when the second and third lethal injection drugs are administered. Dr. Lubarsky testified that while benzodiazepines such as midazolam may, like barbiturate drugs such as sodium thiopental and pentobarbital, induce unconsciousness by inhibiting neuron function, they do so in a materially different way. Id., at 207 . More specifically, Dr. Lubarsky explained that both barbiturates and benzodiazepines initially cause sedation by facilitating the binding of a naturally occurring chemical called gamma-aminobutyric acid (GABA) with GABA receptors, which then impedes the flow of electrical impulses through the neurons in the central nervous system. Id., at 206 . But at higher doses, barbiturates also act as a GABA substitute and mimic its neuron-suppressing effects. Ibid. By contrast, benzodiazepines lack this mimicking function, which means their effect is capped at a lower level of sedation. Ibid. Critically, according to Dr. Lubarsky, this ceiling on midazolam's sedative effect is reached before full anesthesia can be achieved. Ibid . Thus, in his view, while "midazolam unconsciousness is ... sufficient" for "minor procedure[s]," Tr. of Preliminary Injunction Hearing 132-133 (Tr.), it is incapable of keeping someone "insensate and immobile in the face of [more] noxious stimuli," including the extreme pain and discomfort associated with administration of the second and third drugs in Oklahoma's lethal injection protocol, App. 218. Dr. Sasich endorsed Dr. Lubarsky's description of the ceiling effect, and offered similar reasons for reaching the same conclusion. See id., at 243, 248, 262 .

In support of these assertions, both experts cited a variety of evidence. Dr. Lubarsky emphasized, in particular, Arizona's 2014 execution of Joseph Wood, which had been conducted using midazolam and the drug hydromorphone rather than the three-drug cocktail Oklahoma intends to employ. 1 Id., at 176 . Despite being administered 750 milligrams of midazolam, Wood had continued breathing and moving for nearly two hours-which, according to Dr. Lubarsky, would not have occurred "during extremely deep levels of anesthesia." Id., at 177 . Both experts also cited various scientific articles and textbooks to support their conclusions. For instance, *2784 Dr. Lubarsky relied on a study measuring the brain activity of rats that were administered midazolam, which showed that the drug's impact significantly tailed off at higher doses. See Hovinga et al., Pharmacokinetic-EEG Effect Relationship of Midazolam in Aging BN/BiRij Rats, 107 British J. Pharmacology 171, 173, Fig. 2 (1992). He also pointed to a pharmacology textbook that confirmed his description of how benzodiazepines and barbiturates produce their effects, see Stoelting & Hillier 127-128, 140-144, and a survey article concluding that "[m]idazolam cannot be used alone ... to maintain adequate anesthesia," Reves, Fragen, Vinik, & Greenblatt, Midazolam: Pharmacology and Uses, 62 Anesthesiology 310, 318 (1985) (Reves). For his part, Dr. Sasich referred to a separate survey article, which similarly recognized and described the ceiling effect to which benzodiazepines are subject. See Saari, Uusi-Oukari, Ahonen, & Olkkola, Enhancement of GABAergic Activity: Neuropharmacological Effects of Benzodiazepines and Therapeutic Use in Anesthesiology, 63 Pharamacological Rev. 243, 244, 250 (2011) (Saari).

By contrast, Dr. Evans, the State's expert, asserted that a 500-milligram dose of midazolam would "render the person unconscious and 'insensate' during the remainder of the [execution] procedure." App. 294. He rested this conclusion on two interrelated propositions.

First, observing that a therapeutic dose of midazolam to treat anxiety is less than 5 milligrams for a 70-kilogram adult, Dr. Evans emphasized that Oklahoma's planned administration of 500 milligrams of the drug was "at least 100 times the normal therapeutic dose." Ibid. While he acknowledged that "[t]here are no studies that have been done ... administering that much ... midazolam ... to anybody," he noted that deaths had occurred in doses as low as 0.04 to 0.07 milligrams per kilogram (2.8 to 4.9 milligrams for a 70-kilogram adult), and contended that a 500-milligram dose would itself cause death within less than an hour-a conclusion he characterized as "essentially an extrapolation from a toxic effect." Id., at 327 ; see id., at 308 .

Second, in explaining how he reconciled his opinion with the evidence of midazolam's ceiling effect, Dr. Evans testified that while "GABA receptors are found across the entire body," midazolam's ceiling effect is limited to the "spinal cord" and there is "no ceiling effect" at the "higher level of [the] brain." Id., at 311-312 . Consequently, in his view, "as you increase the dose of midazolam, it's a linear effect, so you're going to continue to get an impact from higher doses of the drug," id., at 332 , until eventually "you're paralyzing the brain," id., at 314 . Dr. Evans also understood the chemical source of midazolam's ceiling effect somewhat differently from petitioners' experts. Although he agreed that midazolam produces its effect by "binding to [GABA] receptors," id., at 293 , he appeared to believe that midazolam produced sedation by "inhibiting GABA" from attaching to GABA receptors, not by promoting GABA's sedative effects, id., at 312 . Thus, when asked about Dr. Lubarsky's description of the ceiling effect, Dr. Evans characterized the phenomenon as stemming from "the competitive nature of substances trying to attach to GABA receptors." Id., at 313 .

Dr. Evans cited no scholarly research in support of his opinions. Instead, he appeared to rely primarily on two sources: the Web site www.drugs.com, and a "Material Safety Data Sheet" produced by a midazolam manufacturer. See id., at 303 . Both simply contained general information that covered the experts' areas of agreement.

*2785 D

The District Court denied petitioners' motion for a preliminary injunction. It began by making a series of factual findings regarding the characteristics of midazolam and its use in Oklahoma's execution protocol. Most relevant here, the District Court found that "[t]he proper administration of 500 milligrams of midazolam ... would make it a virtual certainty that an individual will be at a sufficient level of unconsciousness to resist the noxious stimuli which could occur from the application of the second and third drugs." Id., at 77 . Respecting petitioners' contention that there is a "ceiling effect which prevents an increase in dosage from having a corresponding incremental effect on anesthetic depth," the District Court concluded:

"Dr. Evans testified persuasively ... that whatever the ceiling effect of midazolam may be with respect to anesthesia, which takes effect at the spinal cord level, there is no ceiling effect with respect to the ability of a 500 milligram dose of midazolam to effectively paralyze the brain, a phenomenon which is not anesthesia but does have the effect of shutting down respiration and eliminating the individual's awareness of pain." Id., at 78 .

Having made these findings, the District Court held that petitioners had shown no likelihood of success on the merits of their Eighth Amendment claim for two independent reasons. First, it determined that petitioners had "failed to establish that proceeding with [their] execution[s] ... on the basis of the revised protocol presents ... 'an objectively intolerable risk of harm.' " Id., at 96 . Second, the District Court held that petitioners were unlikely to prevail because they had not identified any " 'known and available alternative' " means by which they could be executed-a requirement it understood Baze to impose. Id., at 97 . The District Court concluded that the State "ha[d] affirmatively shown that sodium thiopental and pentobarbital, the only alternatives to which the [petitioners] have even alluded, are not available to the [State]." Id., at 98 .

The Court of Appeals for the Tenth Circuit affirmed. Warner v. Gross, 776 F.3d 721 (2015). It, like the District Court, held that petitioners were unlikely to prevail on the merits because they had failed to prove the existence of " 'known and available alternatives.' " Id., at 732 . "In any event," the court continued, it was unable to conclude that the District Court's factual findings had been clearly erroneous, and thus petitioners had also "failed to establish that the use of midazolam in their executions ... creates a demonstrated risk of severe pain." Ibid.

Petitioners and Charles Warner filed a petition for certiorari and an application to stay their executions. The Court denied the stay application, and Charles Warner was executed on January 15, 2015. See Warner v. Gross, 574 U.S. ----, 135 S.Ct. 824 , 190 L.Ed.2d 903 (2015) (SOTOMAYOR, J., dissenting from denial of certiorari). The Court subsequently granted certiorari and, at the request of the State, stayed petitioners' pending executions.

II

I begin with the second of the Court's two holdings: that the District Court properly found that petitioners did not demonstrate a likelihood of showing that Oklahoma's execution protocol poses an unconstitutional risk of pain. In reaching this conclusion, the Court sweeps aside substantial evidence showing that, while midazolam may be able to induce unconsciousness, it cannot be utilized to maintain unconsciousness in the face of agonizing stimuli. Instead, like the District *2786 Court, the Court finds comfort in Dr. Evans' wholly unsupported claims that 500 milligrams of midazolam will "paralyz[e] the brain." In so holding, the Court disregards an objectively intolerable risk of severe pain.

A

Like the Court, I would review for clear error the District Court's finding that 500 milligrams of midazolam will render someone sufficiently unconscious " 'to resist the noxious stimuli which could occur from the application of the second and third drugs.' " Ante, at 2740 (quoting App. 77). Unlike the Court, however, I would do so without abdicating our duty to examine critically the factual predicates for the District Court's finding-namely, Dr. Evans' testimony that midazolam has a "ceiling effect" only "at the spinal cord level," and that a "500 milligram dose of midazolam" can therefore "effectively paralyze the brain." Id., at 78. To be sure, as the Court observes, such scientific testimony may at times lie at the boundaries of federal courts' expertise. See ante, at 2739 - 2740. But just because a purported expert says something does not make it so. Especially when important constitutional rights are at stake, federal district courts must carefully evaluate the premises and evidence on which scientific conclusions are based, and appellate courts must ensure that the courts below have in fact carefully considered all the evidence presented. Clear error exists "when although there is evidence to support" a finding, "the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed." United States v. United States Gypsum Co., 333 U.S. 364 , 395, 68 S.Ct. 525 , 92 L.Ed. 746 (1948). Here, given the numerous flaws in Dr. Evans' testimony, there can be little doubt that the District Court clearly erred in relying on it.

To begin, Dr. Evans identified no scientific literature to support his opinion regarding midazolam's properties at higher-than-normal doses. Apart from a Material Safety Data Sheet that was relevant only insofar as it suggests that a low dose of midazolam may occasionally be toxic, see ante, at 2745-an issue I discuss further below-Dr. Evans' testimony seems to have been based on the Web site www.drugs.com. The Court may be right that "petitioners do not identify any incorrect statements from drugs.com on which Dr. Evans relied." Ante, at 2745. But that is because there were no statements from drugs.com that supported the critically disputed aspects of Dr. Evans' opinion. If anything, the Web site supported petitioners' contentions, as it expressly cautioned that midazolam "[s]hould not be used alone for maintenance of anesthesia," App. H to Pet. for Cert. 6159, and contained no warning that an excessive dose of midazolam could "paralyze the brain," see id., at 6528-6529.

Most importantly, nothing from drugs.com-or, for that matter, any other source in the record-corroborated Dr. Evans' key testimony that midazolam's ceiling effect is limited to the spinal cord and does not pertain to the brain. Indeed, the State appears to have disavowed Dr. Evans' spinal-cord theory, refraining from even mentioning it in its brief despite the fact that the District Court expressly relied on this testimony as the basis for finding that larger doses of midazolam will have greater anesthetic effects. App. 78. The Court likewise assiduously avoids defending this theory.

That is likely because this aspect of Dr. Evans' testimony was not just unsupported, but was directly refuted by the studies and articles cited by Drs. Lubarsky and Sasich. Both of these experts relied on *2787 academic texts describing benzodiazepines' ceiling effect and explaining why it prevents these drugs from rendering a person completely insensate. See Stoelting & Hillier 141, 144 (describing midazolam's ceiling effect and contrasting the drug with barbiturates); Saari 244 (observing that "abolishment of perception of environmental stimuli cannot usually be generated"). One study further made clear that the ceiling effect is apparent in the brain. See id., at 250.

These scientific sources also appear to demonstrate that Dr. Evans' spinal-cord theory- i.e., that midazolam's ceiling effect is limited to the spinal cord-was premised on a basic misunderstanding of midazolam's mechanism of action. I say "appear" not because the sources themselves are unclear about how midazolam operates: They plainly state that midazolam functions by promoting GABA's inhibitory effects on the central nervous system. See, e.g., Stoelting & Hillier 140. Instead, I use "appear" because discerning the rationale underlying Dr. Evans' testimony is difficult. His spinal-cord theory might, however, be explained at least in part by his apparent belief that rather than promoting GABA's inhibitory effects, midazolam produces sedation by "compet[ing]" with GABA and thus "inhibit[ing]" GABA's effect. App. 312-313. 2 Regardless, I need not delve too deeply into Dr. Evans' alternative scientific reality. It suffices to say that to the extent that Dr. Evans' testimony was based on his understanding of the source of midazolam's pharmacological properties, that understanding was wrong.

These inconsistencies and inaccuracies go to the very heart of Dr. Evans' expert opinion, as they were the key components of his professed belief that one can extrapolate from what is known about midazolam's effect at low doses to conclude that the drug would "paralyz[e] the brain" at Oklahoma's planned dose. Id., at 314. All three experts recognized that there had been no scientific testing on the use of this amount of midazolam in conjunction with these particular lethal injection drugs. See ante, at 2740 - 2741; App. 176 (Lubarsky), 243-244 (Sasich), 327 (Evans). For this reason, as the Court correctly observes, "extrapolation was reasonable." Ante, at 2741. But simply because extrapolation may be reasonable or even required does not mean that every conceivable method of extrapolation can be credited, or that all estimates stemming from purported extrapolation are worthy of belief. Dr. Evans' view was that because 40 milligrams of midazolam could be used to induce unconsciousness, App. 294, and because more drug will generally produce more effect, a significantly larger dose of 500 milligrams would not just induce unconsciousness but allow for its maintenance in the face of extremely painful stimuli, and ultimately *2788 even cause death itself. In his words: "[A]s you increase the dose of midazolam, it's a linear effect, so you're going to continue to get an impact from higher doses of the drug." Id., at 332. If, however, there is a ceiling with respect to midazolam's effect on the brain-as petitioners' experts established there is-then such simplistic logic is not viable. In this context, more is not necessarily better, and Dr. Evans was plainly wrong to presume it would be.

If Dr. Evans had any other basis for the "extrapolation" that led him to conclude 500 milligrams of midazolam would "paralyz[e] the brain," id., at 314, it was even further divorced from scientific evidence and logic. Having emphasized that midazolam had been known to cause approximately 80 deaths, Dr. Evans asserted that his opinion regarding the efficacy of Oklahoma's planned use of the drug represented "essentially an extrapolation from a toxic effect." Id., at 327 (emphasis added); see id., at 308. Thus, Dr. Evans appeared to believe-and again, I say "appeared" because his rationale is not clear-that because midazolam caused some deaths, it would necessarily cause complete unconsciousness and then death at especially high doses. But Dr. Evans also thought, and Dr. Lubarsky confirmed, that these midazolam fatalities had occurred at very low doses-well below what any expert said would produce unconsciousness. See id., at 207, 308. These deaths thus seem to represent the rare, unfortunate side effects that one would expect to see with any drug at normal therapeutic doses; they provide no indication of the effect one would expect midazolam to have on the brain at substantially higher doses. Deaths occur with almost any product. One might as well say that because some people occasionally die from eating one peanut, one hundred peanuts would necessarily induce a coma and death in anyone. 3

In sum, then, Dr. Evans' conclusions were entirely unsupported by any study or third-party source, contradicted by the extrinsic evidence proffered by petitioners, inconsistent with the scientific understanding of midazolam's properties, and apparently premised on basic logical errors. Given these glaring flaws, the District Court's acceptance of Dr. Evans' claim that 500 milligrams of midazolam would "paralyz[e] the brain" cannot be credited. This is not a case "[w]here there are two permissible views of the evidence," and the District Court chose one; rather, it is one where the trial judge credited "one of two or more witnesses" even though that witness failed to tell "a coherent and facially plausible story that is not contradicted by extrinsic evidence." Anderson v. Bessemer City, 470 U.S. 564 , 574-575, 105 S.Ct. 1504 , 84 L.Ed.2d 518 (1985). In other words, this is a case in which the District Court clearly erred. See ibid.

B

Setting aside the District Court's erroneous factual finding that 500 milligrams of midazolam will necessarily "paralyze the brain," the question is whether the Court is nevertheless correct to hold that petitioners *2789 failed to demonstrate that the use of midazolam poses an "objectively intolerable risk" of severe pain. See Baze, 553 U.S., at 50 , 128 S.Ct. 1520 (plurality opinion) (internal quotation marks omitted). I would hold that they made this showing. That is because, in stark contrast to Dr. Evans, petitioners' experts were able to point to objective evidence indicating that midazolam cannot serve as an effective anesthetic that "render[s] a person insensate to pain caused by the second and third [lethal injection] drugs." Ante, at 2743.

As observed above, these experts cited multiple sources supporting the existence of midazolam's ceiling effect. That evidence alone provides ample reason to doubt midazolam's efficacy. Again, to prevail on their claim, petitioners need only establish an intolerable risk of pain, not a certainty. See Baze, 553 U.S., at 50 , 128 S.Ct. 1520 . Here, the State is attempting to use midazolam to produce an effect the drug has never previously been demonstrated to produce, and despite studies indicating that at some point increasing the dose will not actually increase the drug's effect. The State is thus proceeding in the face of a very real risk that the drug will not work in the manner it claims.

Moreover, and perhaps more importantly, the record provides good reason to think this risk is substantial. The Court insists that petitioners failed to provide "probative evidence" as to whether "midazolam's ceiling effect occurs below the level of a 500-milligram dose and at a point at which the drug does not have the effect of rendering a person insensate to pain." Ante, at 2743. It emphasizes that Dr. Lubarsky was unable to say "at what dose the ceiling effect occurs," and could only estimate that it was " '[p]robably after about ... 40 to 50 milligrams.' " Ante, at 2743 (quoting App. 225).

But the precise dose at which midazolam reaches its ceiling effect is irrelevant if there is no dose at which the drug can, in the Court's words, render a person "insensate to pain." Ante, at 2743. On this critical point, Dr. Lubarsky was quite clear. 4 He explained that the drug "does not work to produce" a "lack of consciousness as noxious stimuli are applied," and is "not sufficient to produce a surgical plane of anesthesia in human beings." App. 204. He also noted that "[t]he drug would never be used and has never been used as a sole anesthetic to give anesthesia during a surgery," id., at 223, and asserted that "the drug was not approved by the FDA as a sole anesthetic because after the use of fairly large doses that were sufficient to reach the ceiling effect and produce induction of unconsciousness, the patients responded to the surgery," id., at 219. Thus, Dr. Lubarsky may not have been able to *2790 identify whether this effect would be reached at 40, 50, or 60 milligrams or some higher threshold, but he could specify that at no level would midazolam reliably keep an inmate unconscious once the second and third drugs were delivered. 5

These assertions were amply supported by the evidence of the manner in which midazolam is and can be used. All three experts agreed that midazolam is utilized as the sole sedative only in minor procedures. Dr. Evans, for example, acknowledged that while midazolam may be used as the sole drug in some procedures that are not "terribly invasive," even then "you would [generally] see it used in combination with a narcotic." Id., at 307. And though, as the Court observes, Dr. Sasich believed midazolam could be "used for medical procedures like colonoscopies and gastroscopies," ante, at 2742, he insisted that these procedures were not necessarily painful, and that it would be a "big jump" to conclude that midazolam would be effective to maintain unconsciousness throughout an execution. Tr. 369-370. Indeed, the record provides no reason to think that these procedures cause excruciating pain remotely comparable to that produced by the second and third lethal injection drugs Oklahoma intends to use.

As for more painful procedures, the consensus was also clear: Midazolam is not FDA-approved for, and is not used as, a sole drug to maintain unconsciousness. See App. 171 (Lubarsky), 262 (Sasich), 327 (Evans). One might infer from the fact that midazolam is not used as the sole anesthetic for more serious procedures that it cannot be used for them. But drawing such an inference is unnecessary, as petitioners' experts invoked sources expressly stating as much. In particular, Dr. Lubarsky pointed to a survey article that cited four separate authorities and declared that "[m]idazolam cannot be used alone ... to maintain adequate anesthesia." Reves 318; see also Stoelting & Hillier 145 (explaining that midazolam is used for "induction of anesthesia," and that, " [i]n combination with other drugs, [it] may be used for maintenance of anesthesia" (emphasis added)).

This evidence was alone sufficient, but if one wanted further support for these conclusions it was provided by the Lockett and Wood executions. The procedural flaws that marred the Lockett execution created the conditions for an unintended (and grotesque) experiment on midazolam's efficacy. Due to problems with the IV line, Lockett was not fully paralyzed after the second and third drugs were administered. He had, however, been administered more than enough midazolam to "render an average person unconscious," as the District Court found. App. 57. When Lockett awoke and began to writhe and speak, he demonstrated the critical difference between midazolam's ability to render an inmate unconscious and its ability to maintain the inmate in that state. The Court insists that Lockett's execution involved "only 100 milligrams of midazolam," ante, at 2746, but as *2791 explained previously, more is not necessarily better given midazolam's ceiling effect.

The Wood execution is perhaps even more probative. Despite being given over 750 milligrams of midazolam, Wood gasped and snorted for nearly two hours. These reactions were, according to Dr. Lubarsky, inconsistent with Wood being fully anesthetized, App. 177-178, and belie the claim that a lesser dose of 500 milligrams would somehow suffice. The Court attempts to distinguish the Wood execution on the ground that the timing of Arizona's administration of midazolam was different. Ante, at 2745 - 2746. But as Dr. Lubarsky testified, it did not "matter" whether in Wood's execution the "midazolam was introduced all at once or over ... multiple doses," because "[t]he drug has a sufficient half life that the effect is cumulative." App. 220; see also Saari 253 (midazolam's "elimination half-life ranges from 1.7 to 3.5 h [ours]"). 6 Nor does the fact that Wood's dose of midazolam was paired with hydromorphone rather than a paralytic and potassium chromide, see ante, at 2746, appear to have any relevance-other than that the use of this analgesic drug may have meant that Wood did not experience the same degree of searing pain that an inmate executed under Oklahoma's protocol may face.

By contrast, Florida's use of this same three-drug protocol in 11 executions, see ante, at 2745 - 2746 (citing Brief for State of Florida as Amicus Curiae 1), tells us virtually nothing. Although these executions have featured no obvious mishaps, the key word is "obvious." Because the protocol involves the administration of a powerful paralytic, it is, as Drs. Sasich and Lubarsky explained, impossible to tell whether the condemned inmate in fact remained unconscious. App. 218, 273; see also Baze, 553 U.S., at 71 , 128 S.Ct. 1520 (Stevens, J., concurring in judgment). Even in these executions, moreover, there have been indications of the inmates' possible awareness. See Brief for State of Alabama et al. as Amici Curiae 9-13 (describing the 11 Florida executions, and noting that some allegedly involved blinking and other movement after administration of the three drugs). 7

Finally, none of the State's "safeguards" for administering these drugs would seem to mitigate the substantial risk that midazolam will not work, as the Court contends. See ante, at 2742 - 2743. Protections ensuring that officials have properly secured a viable IV site will not enable midazolam to have an effect that it is chemically incapable of having. Nor is there any indication that the State's monitoring of the inmate's consciousness will be able to anticipate whether the inmate will remain unconscious while the second and third drugs are administered. No one questions whether midazolam can induce unconsciousness. The problem, as Lockett's execution vividly illustrates, is that an unconscious inmate may be awakened by the pain and respiratory distress caused by administration of the second and third *2792 drugs. At that point, even if it were possible to determine whether the inmate is conscious-dubious, given the use of a paralytic-it is already too late. Presumably for these reasons, the Tenth Circuit characterized the District Court's reliance on these procedural mechanisms as "not relevant to its rejection of [petitioners'] claims regarding the inherent characteristics of midazolam." Warner, 776 F.3d, at 733 .

C

The Court not only disregards this record evidence of midazolam's inadequacy, but also fails to fully appreciate the procedural posture in which this case arises. Petitioners have not been accorded a full hearing on the merits of their claim. They were granted only an abbreviated evidentiary proceeding that began less than three months after the State issued its amended execution protocol; they did not even have the opportunity to present rebuttal evidence after Dr. Evans testified. They sought a preliminary injunction, and thus were not required to prove their claim, but only to show that they were likely to succeed on the merits. See Winter v. Natural Resources Defense Council, Inc., 555 U.S. 7 , 20, 129 S.Ct. 365 , 172 L.Ed.2d 249 (2008); Hill v. McDonough, 547 U.S. 573 , 584, 126 S.Ct. 2096 , 165 L.Ed.2d 44 (2006).

Perhaps the State could prevail after a full hearing, though this would require more than Dr. Evans' unsupported testimony. At the preliminary injunction stage, however, petitioners presented compelling evidence suggesting that midazolam will not work as the State intends. The State, by contrast, offered absolutely no contrary evidence worth crediting. Petitioners are thus at the very least likely to prove that, due to midazolam's inherent deficiencies, there is a constitutionally intolerable risk that they will be awake, yet unable to move, while chemicals known to cause "excruciating pain" course through their veins. Baze, 553 U.S., at 71 , 128 S.Ct. 1520 (Stevens, J., concurring in judgment).

III

The Court's determination that the use of midazolam poses no objectively intolerable risk of severe pain is factually wrong. The Court's conclusion that petitioners' challenge also fails because they identified no available alternative means by which the State may kill them is legally indefensible.

A

This Court has long recognized that certain methods of execution are categorically off-limits. The Court first confronted an Eighth Amendment challenge to a method of execution in Wilkerson v. Utah, 99 U.S. 130 , 25 L.Ed. 345 (1879). Although Wilkerson approved the particular method at issue-the firing squad-it made clear that "public dissection," "burning alive," and other "punishments of torture ... in the same line of unnecessary cruelty, are forbidden by [the Eighth A]mendment to the Constitution." Id., at 135-136 . Eleven years later, in rejecting a challenge to the first proposed use of the electric chair, the Court again reiterated that "if the punishment prescribed for an offense against the laws of the State were manifestly cruel and unusual, as burning at the stake, crucifixion, breaking on the wheel, or the like, it would be the duty of the courts to adjudge such penalties to be within the constitutional prohibition." In re Kemmler, 136 U.S. 436 , 446, 10 S.Ct. 930 , 34 L.Ed. 519 (1890).

In the more than a century since, the Members of this Court have often had cause to debate the full scope of the Eighth Amendment's prohibition of cruel *2793 and unusual punishment. See, e.g., Furman v. Georgia, 408 U.S. 238 , 92 S.Ct. 2726 , 33 L.Ed.2d 346 (1972). But there has been little dispute that it at the very least precludes the imposition of "barbarous physical punishments." Rhodes v. Chapman, 452 U.S. 337 , 345, 101 S.Ct. 2392 , 69 L.Ed.2d 59 (1981); see, e.g., Solem v. Helm, 463 U.S. 277 , 284, 103 S.Ct. 3001 , 77 L.Ed.2d 637 (1983); id., at 312-313 , 103 S.Ct. 3001 (Burger, C.J., dissenting); Baze, 553 U.S., at 97-99 , 128 S.Ct. 1520 (THOMAS, J., concurring in judgment); Harmelin v. Michigan, 501 U.S. 957 , 976, 111 S.Ct. 2680 , 115 L.Ed.2d 836 (1991) (opinion of SCALIA, J.). Nor has there been any question that the Amendment prohibits such "inherently barbaric punishments under all circumstances ." Graham v. Florida, 560 U.S. 48 , 59, 130 S.Ct. 2011 , 176 L.Ed.2d 825 (2010) (emphasis added). Simply stated, the "Eighth Amendment categorically prohibits the infliction of cruel and unusual punishments." Penry v. Lynaugh, 492 U.S. 302 , 330, 109 S.Ct. 2934 , 106 L.Ed.2d 256 (1989) (emphasis added).

B

The Court today, however, would convert this categorical prohibition into a conditional one. A method of execution that is intolerably painful-even to the point of being the chemical equivalent of burning alive-will, the Court holds, be unconstitutional if, and only if, there is a "known and available alternative" method of execution. Ante, at 2762 - 2763. It deems Baze to foreclose any argument to the contrary. Ante, at 2762.

Baze held no such thing. In the first place, the Court cites only the plurality opinion in Baze as support for its known-and-available-alternative requirement. See ibid. Even assuming that the Baze plurality set forth such a requirement-which it did not-none of the Members of the Court whose concurrences were necessary to sustain the Baze Court's judgment articulated a similar view. See 553 U.S., at 71-77, 87 , 128 S.Ct. 1520 (Stevens, J., concurring in judgment); id., at 94, 99-107 , 128 S.Ct. 1520 (THOMAS, J., concurring in judgment); id., at 107-108, 113 , 128 S.Ct. 1520 (BREYER, J., concurring in judgment). In general, "the holding of the Court may be viewed as that position taken by those Members who concurred in the judgments on the narrowest grounds." Marks v. United States, 430 U.S. 188 , 193, 97 S.Ct. 990 , 51 L.Ed.2d 260 (1977) (internal quotation marks omitted). And as the Court observes, ante, at 2738, n. 2, the opinion of Justice THOMAS, joined by Justice SCALIA, took the broadest position with respect to the degree of intent that state officials must have in order to have violated the Eighth Amendment, concluding that only a method of execution deliberately designed to inflict pain, and not one simply designed with deliberate indifference to the risk of severe pain, would be unconstitutional. 553 U.S., at 94 , 128 S.Ct. 1520 (THOMAS, J., concurring in judgment). But this understanding of the Eighth Amendment's intent requirement is unrelated to, and thus not any broader or narrower than, the requirement the Court now divines from Baze . Because the position that a plaintiff challenging a method of execution under the Eighth Amendment must prove the availability of an alternative means of execution did not "represent the views of a majority of the Court," it was not the holding of the Baze Court. CTS Corp. v. Dynamics Corp. of America, 481 U.S. 69 , 81, 107 S.Ct. 1637 , 95 L.Ed.2d 67 (1987).

In any event, even the Baze plurality opinion provides no support for the Court's proposition. To be sure, that opinion contains the following sentence: "[The condemned] must show that the risk is substantial *2794 when compared to the known and available alternatives." 553 U.S., at 61 , 128 S.Ct. 1520 . But the meaning of that key sentence and the limits of the requirement it imposed are made clear by the sentence directly preceding it: "A stay of execution may not be granted on grounds such as those asserted here unless the condemned prisoner establishes that the State's lethal injection protocol creates a demonstrated risk of severe pain." Ibid. (emphasis added). In Baze, the very premise of the petitioners' Eighth Amendment claim was that they had "identified a significant risk of harm [in Kentucky's protocol] that [could] be eliminated by adopting alternative procedures." Id., at 51 , 128 S.Ct. 1520 . Their basic theory was that even if the risk of pain was only, say, 25%, that risk would be objectively intolerable if there was an obvious alternative that would reduce the risk to 5%. See Brief for Petitioners in Baze v. Rees, O.T. 2007, No. 07-5439, p. 29 ("In view of the severity of the pain risked and the ease with which it could be avoided, Petitioners should not have been required to show a high likelihood that they would suffer such pain ..."). Thus, the "grounds ... asserted" for relief in Baze were that the State's protocol was intolerably risky given the alternative procedures the State could have employed.

Addressing this claim, the Baze plurality clarified that "a condemned prisoner cannot successfully challenge a State's method of execution merely by showing a slightly or marginally safer alternative," 553 U.S., at 51 , 128 S.Ct. 1520 ; instead, to succeed in a challenge of this type, the comparative risk must be "substantial," id., at 61 , 128 S.Ct. 1520 . Nowhere did the plurality suggest that all challenges to a State's method of execution would require this sort of comparative-risk analysis. Recognizing the relevance of available alternatives is not at all the same as concluding that their absence precludes a claimant from showing that a chosen method carries objectively intolerable risks. If, for example, prison officials chose a method of execution that has a 99% chance of causing lingering and excruciating pain, certainly that risk would be objectively intolerable whether or not the officials ignored other methods in making this choice. Irrespective of the existence of alternatives, there are some risks "so grave that it violates contemporary standards of decency to expose anyone unwillingly to" them. Helling v. McKinney, 509 U.S. 25 , 36, 113 S.Ct. 2475 , 125 L.Ed.2d 22 (1993) (emphasis in original).

That the Baze plurality's statement regarding a condemned inmate's ability to point to an available alternative means of execution pertained only to challenges premised on the existence of such alternatives is further evidenced by the opinion's failure to distinguish or even mention the Court's unanimous decision in Hill v. McDonough, 547 U.S. 573 , 126 S.Ct. 2096 , 165 L.Ed.2d 44 . Hill held that a § 1983 plaintiff challenging a State's method of execution need not "identif[y] an alternative, authorized method of execution." Id., at 582 , 126 S.Ct. 2096 . True, as the Court notes, ante, at 2738 - 2739, Hill did so in the context of addressing § 1983' s pleading standard, rejecting the proposed alternative-means requirement because the Court saw no basis for the "[i]mposition of heightened pleading requirements." 547 U.S., at 582 , 126 S.Ct. 2096 . But that only confirms that the Court in Hill did not view the availability of an alternative means of execution as an element of an Eighth Amendment claim: If it had, then requiring the plaintiff to plead this element would not have meant imposing a heightened standard at all, but rather would have been entirely consistent with "traditional pleading requirements."

*2795 Ibid. ; see Ashcroft v. Iqbal, 556 U.S. 662 , 678, 129 S.Ct. 1937 , 173 L.Ed.2d 868 (2009). The Baze plurality opinion should not be understood to have so carelessly tossed aside Hill 's underlying premise less than two years later.

C

In reengineering Baze to support its newfound rule, the Court appears to rely on a flawed syllogism. If the death penalty is constitutional, the Court reasons, then there must be a means of accomplishing it, and thus some available method of execution must be constitutional. See ante, at 2732 - 2733, 2738 - 2739. But even accepting that the death penalty is, in the abstract, consistent with evolving standards of decency, but see ante, p. 2760 - 2761 (BREYER, J., dissenting), the Court's conclusion does not follow. The constitutionality of the death penalty may inform our conception of the degree of pain that would render a particular method of imposing it unconstitutional. See Baze, 553 U.S., at 47 , 128 S.Ct. 1520 (plurality opinion) (because "[s]ome risk of pain is inherent in any method of execution," "[i]t is clear ... the Constitution does not demand the avoidance of all risk of pain"). But a method of execution that is "barbarous," Rhodes, 452 U.S., at 345 , 101 S.Ct. 2392 , or "involve[s] torture or a lingering death," Kemmler, 136 U.S., at 447 , 10 S.Ct. 930 , does not become less so just because it is the only method currently available to a State. If all available means of conducting an execution constitute cruel and unusual punishment, then conducting the execution will constitute cruel and usual punishment. Nothing compels a State to perform an execution. It does not get a constitutional free pass simply because it desires to deliver the ultimate penalty; its ends do not justify any and all means. If a State wishes to carry out an execution, it must do so subject to the constraints that our Constitution imposes on it, including the obligation to ensure that its chosen method is not cruel and unusual. Certainly the condemned has no duty to devise or pick a constitutional instrument of his or her own death.

For these reasons, the Court's available-alternative requirement leads to patently absurd consequences. Petitioners contend that Oklahoma's current protocol is a barbarous method of punishment-the chemical equivalent of being burned alive. But under the Court's new rule, it would not matter whether the State intended to use midazolam, or instead to have petitioners drawn and quartered, slowly tortured to death, or actually burned at the stake: because petitioners failed to prove the availability of sodium thiopental or pentobarbital, the State could execute them using whatever means it designated. But see Baze, 553 U.S., at 101-102 , 128 S.Ct. 1520 (THOMAS, J., concurring in judgment) ("It strains credulity to suggest that the defining characteristic of burning at the stake, disemboweling, drawing and quartering, beheading, and the like was that they involved risks of pain that could be eliminated by using alternative methods of execution"). 8 The Eighth Amendment cannot possibly countenance such a result.

D

In concocting this additional requirement, the Court is motivated by a desire to preserve States' ability to conduct executions *2796 in the face of changing circumstances. See ante, at 2732 - 2734, 2745 - 2746. It is true, as the Court details, that States have faced "practical obstacle[s]" to obtaining lethal injection drugs since Baze was decided. Ante, at 2732 - 2733. One study concluded that recent years have seen States change their protocols "with a frequency that is unprecedented among execution methods in this country's history." Denno, Lethal Injection Chaos Post- Baze, 102 Geo. L. J. 1331 , 1335 (2014).

But why such developments compel the Court's imposition of further burdens on those facing execution is a mystery. Petitioners here had no part in creating the shortage of execution drugs; it is odd to punish them for the actions of pharmaceutical companies and others who seek to disassociate themselves from the death penalty-actions which are, of course, wholly lawful. Nor, certainly, should these rapidly changing circumstances give us any greater confidence that the execution methods ultimately selected will be sufficiently humane to satisfy the Eighth Amendment. Quite the contrary. The execution protocols States hurriedly devise as they scramble to locate new and untested drugs, see supra, at 2781 - 2782, are all the more likely to be cruel and unusual-presumably, these drugs would have been the States' first choice were they in fact more effective. But see Denno, The Lethal Injection Quandry: How Medicine Has Dismantled the Death Penalty, 76 Ford. L. Rev. 49, 65-79 (2007) (describing the hurried and unreasoned process by which States first adopted the original three-drug protocol). Courts' review of execution methods should be more, not less, searching when States are engaged in what is in effect human experimentation.

It is also worth noting that some condemned inmates may read the Court's surreal requirement that they identify the means of their death as an invitation to propose methods of executions less consistent with modern sensibilities. Petitioners here failed to meet the Court's new test because of their assumption that the alternative drugs to which they pointed, pentobarbital and sodium thiopental, were available to the State. See ante, at 2737 - 2738. This was perhaps a reasonable assumption, especially given that neighboring Texas and Missouri still to this day continue to use pentobarbital in executions. See The Death Penalty Institute, Execution List 2015, online at www.deathpenaltyinfo.org/execution-list-2015 (as visited June 26, 2015, and available in the Clerk of the Court's case file).

In the future, however, condemned inmates might well decline to accept States' current reliance on lethal injection. In particular, some inmates may suggest the firing squad as an alternative. Since the 1920's, only Utah has utilized this method of execution. See S. Banner, The Death Penalty 203 (2002); Johnson, Double Murderer Executed by Firing Squad in Utah, N.Y. Times, June 19, 2010, p. A12. But there is evidence to suggest that the firing squad is significantly more reliable than other methods, including lethal injection using the various combinations of drugs thus far developed. See A. Sarat, Gruesome Spectacles: Botched Executions and America's Death Penalty, App. A, p. 177 (2014) (calculating that while 7.12% of the 1,054 executions by lethal injection between 1900 and 2010 were "botched," none of the 34 executions by firing squad had been). Just as important, there is some reason to think that it is relatively quick and painless. See Banner, supra, at 203.

Certainly, use of the firing squad could be seen as a devolution to a more primitive era. See Wood v. Ryan, 759 F.3d 1076 , 1103 (C.A.9 2014) (Kozinski, C.J., dissenting from denial of rehearing en banc).

*2797 That is not to say, of course, that it would therefore be unconstitutional. But lethal injection represents just the latest iteration of the States' centuries-long search for "neat and non-disfiguring homicidal methods." C. Brandon, The Electric Chair: An Unnatural American History 39 (1999) (quoting Editorial, New York Herald, Aug. 10, 1884); see generally Banner, supra, at 169-207. A return to the firing squad-and the blood and physical violence that comes with it-is a step in the opposite direction. And some might argue that the visible brutality of such a death could conceivably give rise to its own Eighth Amendment concerns. See Campbell v. Wood, 511 U.S. 1119 , 1121-1123, 114 S.Ct. 2125 , 128 L.Ed.2d 682 (1994) (Blackmun, J., dissenting from denial of stay of execution and certiorari); Glass v. Louisiana, 471 U.S. 1080 , 1085, 105 S.Ct. 2159 , 85 L.Ed.2d 514 (1985) (Brennan, J., dissenting from denial of certiorari). At least from a condemned inmate's perspective, however, such visible yet relatively painless violence may be vastly preferable to an excruciatingly painful death hidden behind a veneer of medication. The States may well be reluctant to pull back the curtain for fear of how the rest of us might react to what we see. But we deserve to know the price of our collective comfort before we blindly allow a State to make condemned inmates pay it in our names.

* * *

"By protecting even those convicted of heinous crimes, the Eighth Amendment reaffirms the duty of the government to respect the dignity of all persons." Roper v. Simmons, 543 U.S. 551 , 560, 125 S.Ct. 1183 , 161 L.Ed.2d 1 (2005). Today, however, the Court absolves the State of Oklahoma of this duty. It does so by misconstruing and ignoring the record evidence regarding the constitutional insufficiency of midazolam as a sedative in a three-drug lethal injection cocktail, and by imposing a wholly unprecedented obligation on the condemned inmate to identify an available means for his or her own execution. The contortions necessary to save this particular lethal injection protocol are not worth the price. I dissent.

--------

FootNotes

^* The syllabus constitutes no part of the opinion of the Court but has been prepared by the Reporter of Decisions for the convenience of the reader. See United States v. Detroit Timber & Lumber Co., 200 U.S. 321 , 337, 26 S.Ct. 282 , 50 L.Ed. 499 .

¹ The three other drug combinations that Oklahoma may administer are: (1) a single dose of pentobarbital, (2) a single dose of sodium thiopental, and (3) a dose of midazolam followed by a dose of hydromorphone.

² Justice SOTOMAYOR's dissent (hereinafter principal dissent), post, at 2793 - 2794, inexplicably refuses to recognize that THE CHIEF JUSTICE's opinion in Baze sets out the holding of the case. In Baze, the opinion of THE CHIEF JUSTICE was joined by two other Justices. Justices SCALIA and THOMAS took the broader position that a method of execution is consistent with the Eighth Amendment unless it is deliberately designed to inflict pain. 553 U.S., at 94 , 128 S.Ct. 1520 (THOMAS, J. concurring in judgment). Thus, as explained in Marks v. United States, 430 U.S. 188 , 193, 97 S.Ct. 990 , 51 L.Ed.2d 260 (1977), THE CHIEF JUSTICE's opinion sets out the holding of the case. It is for this reason that petitioners base their argument on the rule set out in that opinion. See Brief for Petitioners 25, 28.

³ Drs. Lubarsky and Sasich, petitioners' key witnesses, both testified that midazolam is inappropriate for a third reason, namely, that it creates a risk of "paradoxical reactions" such as agitation, hyperactivity, and combativeness. App. 175 (expert report of Dr. Lubarsky); id., at 242, 244 (expert report of Dr. Sasich). The District Court found, however, that the frequency with which a paradoxical reaction occurs "is speculative" and that the risk "occurs with the highest frequency in low therapeutic doses." Id., at 78. Indeed, Dr. Sasich conceded that the incidence or risk of paradoxical reactions with midazolam "is unknown" and that reports estimate the risk to vary only "from 1% to above 10%." Id., at 244. Moreover, the mere fact that a method of execution might result in some unintended side effects does not amount to an Eighth Amendment violation. "[T]he Constitution does not demand the avoidance of all risk of pain." Baze, 553 U.S., at 47 , 128 S.Ct. 1520 (plurality opinion).

⁴ The principal dissent misunderstands the record when it bizarrely suggests that midazolam is about as dangerous as a peanut. Post, at 2788. Dr. Evans and Dr. Lubarsky agreed that midazolam has caused fatalities in doses as low as 0.04 to 0.07 milligrams per kilogram. App. 217, 294. Even if death from such low doses is a "rare, unfortunate side effec[t]," post, at 2788, the District Court found that a massive 500-milligram dose-many times higher than the lowest dose reported to have produced death-will likely cause death in under an hour. App. 76-77.

⁵ Petitioners' experts also declined to testify that a 500-milligram dose of midazolam is always insufficient to place a person in a coma and render him insensate to pain. Dr. Lubarsky argued only that the 500-milligram dose cannot "reliably" produce a coma. Id., 228. And when Dr. Sasich was asked whether he could say to a reasonable degree of certainty that a 500-milligram dose of midazolam would not render someone unconscious, he replied that he could not. Id., at 271-272. A product label for midazolam that Dr. Sasich attached to his expert report also acknowledged that an overdose of midazolam can cause a coma. See Expert Report of Larry D. Sasich, in No. 14-6244 (CA10), p. 34.

⁶ The principal dissent emphasizes Dr. Lubarsky's supposedly contrary testimony, but the District Court was entitled to credit Dr. Evans (and Dr. Sasich) instead of Dr. Lubarsky on this point. And the District Court had strong reasons not to credit Dr. Lubarsky, who even argued that a protocol that includes sodium thiopental is "constructed to produce egregious harm and suffering." App. 227.

⁷ GABA is "an amino acid that functions as an inhibitory neurotransmitter in the brain and spinal cord." Mosby's Medical Dictionary 782 (7th ed. 2006).

⁸ The principal dissent emphasizes Dr. Lubarsky's testimony that it is irrelevant that Wood was administered the drug over a 2-hour period. Post, at 2790 - 2791. But Dr. Evans disagreed and testified that if a 750-milligram dose "was spread out over a long period of time," such as one hour ( i.e., half the time at issue in the Wood execution), the drug might not be as effective as if it were administered all at once. Tr. 667. The principal dissent states that this "pronouncement was entirely unsupported," post, at 2791, n. 6, but it was supported by Dr. Evans' expertise and decades of experience. And it would be unusual for an expert testifying on the stand to punctuate each sentence with citation to a medical journal.After the Wood execution, Arizona commissioned an independent assessment of its execution protocol and the Wood execution. According to that report, the IV team leader, medical examiner, and an independent physician all agreed that the dosage of midazolam "would result in heavy sedation." Ariz. Dept. of Corrections, Assessment and Review of the Ariz. Dept. of Corrections Execution Protocols 46, 48 (Dec. 15, 2014), online at https:// corrections.az.gov/sites/default/files/documents/PDFs/arizona_final_report_ 12_15_14_w_cover.pdf. And far from blaming midazolam for the Wood execution, the report recommended that Arizona replace its two-drug protocol with Oklahoma's three-drug protocol that includes a 500-milligram dose of midazolam as the first drug. Id., at 49.

¹ Generally: Baze v. Rees, 553 U.S. 35 , 94-97, 128 S.Ct. 1520 , 170 L.Ed.2d 420 (2008) (THOMAS, J., concurring in judgment) (explaining that the Cruel and Unusual Punishments Clause does not prohibit the death penalty, but only torturous punishments); Graham v. Collins, 506 U.S. 461 , 488, 113 S.Ct. 892 , 122 L.Ed.2d 260 (1993) (THOMAS, J., concurring); Gardner v. Florida, 430 U.S. 349 , 371, 97 S.Ct. 1197 , 51 L.Ed.2d 393 (1977) (Rehnquist, J., dissenting) ("The prohibition of the Eighth Amendment relates to the character of the punishment, and not to the process by which it is imposed"). On reliability: Kansas v. Marsh, 548 U.S. 163 , 181, 126 S.Ct. 2516 , 165 L.Ed.2d 429 (2006) (noting that the death penalty remains constitutional despite imperfections in the criminal justice system); McGautha v. California, 402 U.S. 183 , 221, 91 S.Ct. 1454 , 28 L.Ed.2d 711 (1971) ("[T]he Federal Constitution, which marks the limits of our authority in these cases, does not guarantee trial procedures that are the best of all worlds, or that accord with the most enlightened ideas of students of the infant science of criminology, or even those that measure up to the individual predilections of members of this Court"). On arbitrariness: Ring v. Arizona, 536 U.S. 584 , 610, 122 S.Ct. 2428 , 153 L.Ed.2d 556 (2002) (SCALIA, J., concurring) (explaining that what compelled States to specify " 'aggravating factors' " designed to limit the death penalty to the worst of the worst was this Court's baseless jurisprudence concerning juror discretion); McCleskey v. Kemp, 481 U.S. 279 , 308-312, 107 S.Ct. 1756 , 95 L.Ed.2d 262 (1987) (noting that various procedures, including the right to a jury trial, constitute a defendant's protection against arbitrariness in the application of the death penalty). On excessive delays: Knight v. Florida, 528 U.S. 990 , 120 S.Ct. 459 , 145 L.Ed.2d 370 (1999) (THOMAS, J., concurring in denial of certiorari) ("I am unaware of any support in the American constitutional tradition or in this Court's precedent for the proposition that a defendant can avail himself of the panoply of appellate and collateral procedures and then complain when his execution is delayed"); see also Johnson v. Bredesen, 558 U.S. 1067 , 1070, 130 S.Ct. 541 , 175 L.Ed.2d 552 (2009) (THOMAS, J., concurring in denial of certiorari). And on the decline in use of the death penalty: Atkins v. Virginia, 536 U.S. 304 , 345, 122 S.Ct. 2242 , 153 L.Ed.2d 335 (2002) (SCALIA, J., dissenting); Woodson v. North Carolina, 428 U.S. 280 , 308-310, 96 S.Ct. 2978 , 49 L.Ed.2d 944 (1976) (Rehnquist, J., dissenting).

² For some, a faith in the jury seems to be correlated to that institution's likelihood of preventing imposition of the death penalty. See, e.g., Ring v. Arizona, 536 U.S. 584 , 614, 122 S.Ct. 2428 , 153 L.Ed.2d 556 (2002) (BREYER, J., concurring in judgment) (arguing that "the Eighth Amendment requires that a jury, not a judge, make the decision to sentence a defendant to death"); Wainwright v. Witt, 469 U.S. 412 , 440, n. 1, 105 S.Ct. 844 , 83 L.Ed.2d 841 (1985) (Brennan, J., dissenting) ("However heinous Witt's crime, the majority's vivid portrait of its gruesome details has no bearing on the issue before us. It is not for this Court to decide whether Witt deserves to die. That decision must first be made by a jury of his peers").

³ For his part, Justice BREYER explains that his experience on the Court has shown him "discrepancies for which [he] can find no rational explanations." Post, at 2763. Why, he asks, did one man receive death for a single-victim murder, while another received life for murdering a young mother and nearly killing her infant? Ibid. The outcomes in those two cases may not be morally compelled, but there was certainly a rational explanation for them: The first man, who had previously confessed to another murder, killed a disabled man who had offered him a place to stay for the night. State v. Badgett, 361 N.C. 234 , 239-240, 644 S.E.2d 206 , 209-210 (2007). The killer stabbed his victim's throat and prevented him from seeking medical attention until he bled to death. Ibid. The second man expressed remorse for his crimes and claimed to suffer from mental disorders. See Charbonneau, Andre Edwards Sentenced to Life in Prison for 2001 Murder, WRAL, Mar. 26, 2004, online at http://www.wral. com/news/local/story/109648 (all Internet materials as visited June 25, 2015, and available in Clerk of Court's case file); Charbonneau, Jury Finds Andre Edwards Guilty of First-Degree Murder, WRAL, Mar. 23, 2004, online at http://www.wral.com/news/local/story/109563. The other "discrepancies" similarly have "rational" explanations, even if reasonable juries could have reached different results.

⁴ Justice BREYER appears to acknowledge that our decision holding mandatory death penalty schemes unconstitutional, Woodson v. North Carolina, 428 U.S. 280 , 96 S.Ct. 2978 , 49 L.Ed.2d 944 (1976) (plurality opinion), may have introduced the problem of arbitrary application. Post, at 2762. I agree that Woodson eliminated one reliable legislative response to concerns about arbitrariness. Graham v. Collins, 506 U.S. 461 , 486, 113 S.Ct. 892 , 122 L.Ed.2d 260 (1993) (THOMAS, J., concurring). Because that decision was also questionable on constitutional grounds, id., at 486-488 , 113 S.Ct. 892 , I would be willing to revisit it in a future case.

¹ Hydromorphone is a powerful analgesic similar to morphine or heroin. See R. Stoelting & S. Hillier, Pharmacology & Physiology in Anesthetic Practice 87-88 (4th ed. 2006) (Stoelting & Hillier).

² The Court disputes this characterization of Dr. Evans' testimony, insisting that Dr. Evans accurately described midazolam's properties in the written report he submitted prior to the hearing below, and suggesting that petitioners' experts would have "dispute[d] the accuracy" of this explanation were it in fact wrong. Ante, at 2744. But Dr. Evans' written report simply said midazolam "produces different levels of central nervous system (CNS) depression through binding to [GABA] receptors." App. 293. That much is true. Only after Drs. Sasich and Lubarsky testified did Dr. Evans further claim that midazolam produced CNS depression by binding to GABA receptors and thereby preventing GABA itself from binding to those receptors -which is where he went wrong. The Court's further observation that Dr. Lubarsky also used a variant on the word "inhibiting" in his testimony-in saying that GABA's " ' inhibition of brain activity is accentuated by midazolam,' " ante, at 2744 (quoting App. 232)-is completely nonresponsive. "Inhibiting" is a perfectly good word; the problem here is the manner in which Dr. Evans used it in a sentence.

³ For all the reasons discussed in Part II-B, infra, and contrary to the Court's claim, see ante, at 2741 - 2742, n. 4, there are good reasons to doubt that 500 milligrams of midazolam will, in light of the ceiling effect, inevitably kill someone. The closest the record comes to providing support for this contention is the fleeting mention in the FDA-approved product label that one of the possible consequences of midazolam overdosage is coma. See ante, at 2742, n. 5. Moreover, even if this amount of the drug could kill some people in "under an hour," ante, at 2742, n. 4, that would not necessarily mean that the condemned would be insensate during the approximately 10 minutes it takes for the paralytic and potassium chloride to do their work.

⁴ Dr. Sasich, as the Court emphasizes, was perhaps more hesitant to reach definitive conclusions, see ante, at 2742, and n. 5, 2743 - 2744, but the statements highlighted by the Court largely reflect his (truthful) observations that no testing has been done at doses of 500 milligrams, and his inability to pinpoint the precise dose at which midazolam's ceiling effect might be reached. Dr. Sasich did not, as the Court suggests, claim that midazolam's ceiling effect would be reached only after a person became fully insensate to pain. Ante, at 2743 - 2744. What Dr. Sasich actually said was: "As the dose increases, the benzodiazepines are expected to produce sedation, amnesia, and finally lack of response to stimuli such as pain (unconsciousness)." App. 243. In context, it is clear that Dr. Sasich was simply explaining that a drug like midazolam can be used to induce unconsciousness-an issue that was and remains undisputed-not that it could render an inmate sufficiently unconscious to resist all noxious stimuli. Indeed, it was midazolam's possible inability to serve the latter function that led Dr. Sasich to conclude that "it is not an appropriate drug to use when administering a paralytic followed by potassium chloride." Id., at 248.

⁵ The Court claims that the District Court could have properly disregarded Dr. Lubarsky's testimony because he asserted that a protocol with sodium thiopental would " 'produce egregious harm and suffering.' " Ante, at 2743, n. 6 (quoting App. 227). But Dr. Lubarsky did not testify that, like midazolam, sodium thiopental would not render an inmate fully insensate even if properly administered; rather, he simply observed that he had previously contended that protocols using that drug were ineffective. See App. 227. He was presumably referring to an article he coauthored that found many condemned inmates were not being successfully delivered the dose of sodium thiopental necessary to fully anesthetize them. See Baze, 553 U.S., at 67 , 128 S.Ct. 1520 (ALITO, J., concurring) (discussing this study).

⁶ The Court asserts that the State refuted these contentions, pointing to Dr. Evans' testimony that 750 milligrams of the drug "might not have the effect that was sought" if administered over an hour. Tr. 667; see ante, at 2745 - 2746, n. 6. But as has been the theme here, this pronouncement was entirely unsupported, and appears to be contradicted by the secondary sources cited by petitioners' experts.

⁷ The fact that courts in Florida have approved the use of midazolam in this fashion is arguably slightly more relevant, though it is worth noting that the majority of these decisions were handed down before the Lockett and Wood executions, and that some relied, as here, on Dr. Evans' testimony. See ante, at 2739 - 2740.

⁸ The Court protests that its holding does not extend so far, deriding this description of the logical implications of its legal rule as "simply not true" and "outlandish rhetoric." Ante, at 2746. But presumably when the Court imposes a "requirement o[n] all Eighth Amendment method-of-execution claims," that requirement in fact applies to " all " methods of execution, without exception. Ante, at 2731 (emphasis added).

--------