Erlinger v. United States

602 U.S. 821

• Court: Supreme Court of the United States
• Decided: June 21, 2024
• Docket: 23-370
• Status: Published

Opinion

PRELIMINARY PRINT

Volume 602 U. S. Part 1 Pages 821–898

OFFICIAL REPORTS OF

THE SUPREME COURT June 21, 2024

REBECCA A. WOMELDORF reporter of decisions

NOTICE: This preliminary print is subject to formal revision before the bound volume is published. Users are requested to notify the Reporter of Decisions, Supreme Court of the United States, Washington, D.C. 20543, pio@supremecourt.gov, of any typographical or other formal errors. OCTOBER TERM, 2023 821

Syllabus

ERLINGER v. UNITED STATES certiorari to the united states court of appeals for the seventh circuit No. 23–370. Argued March 27, 2024—Decided June 21, 2024 Paul Erlinger pleaded guilty to being a felon in possession of a frearm in violation of 18 U. S. C. § 922(g). At sentencing, the judge found Mr. Er- linger eligible for an enhanced sentence under the Armed Career Crimi- nal Act, § 924(e)(1), which increases the penalty for a § 922(g) conviction from a maximum sentence of 10 years to a mandatory minimum sentence of 15 years when the defendant has three or more qualifying convictions for offenses committed on different occasions. Subsequently, the Sev- enth Circuit held in unrelated decisions that two of the offenses on which the government relied for Mr. Erlinger's sentence enhancement no longer qualifed as ACCA predicate offenses. The District Court va- cated Mr. Erlinger's sentence and scheduled resentencing. At the re- sentencing hearing, prosecutors again pursued an ACCA sentence en- hancement based on a new set of 26-year-old convictions for burglaries

committed by Mr. Erlinger over the course of several days. Mr. Er- linger protested that the burglaries were part of a single criminal epi- sode and did not occur on separate occasions, as required by ACCA. Moreover, Mr. Erlinger argued that the question whether he committed these prior burglaries during a single episode or on distinct occasions required an assessment of the facts surrounding those offenses, and that the Fifth and Sixth Amendments required that a jury make that assess- ment. The District Court rejected Mr. Erlinger's request for a jury and issued a 15-year enhanced sentence. On appeal, the government confessed error. Pointing to this Court's recent decision in Wooden v. United States, 595 U. S. 360, which acknowledged that an ACCA “occa- sions inquiry” can be intensely factual in nature, the government ad- mitted that given the factual nature of the inquiry and its impact on a defendant's sentence, the Constitution requires a jury to decide unani- mously and beyond a reasonable doubt whether Mr. Erlinger's prior offenses were committed on different occasions. This Court granted certiorari and appointed counsel to defend the judgment below. Held: The Fifth and Sixth Amendments require a unanimous jury to make the determination beyond a reasonable doubt that a defendant's past offenses were committed on separate occasions for ACCA purposes. Pp. 829–849. 822 ERLINGER v. UNITED STATES

(a) The Sixth Amendment promises that “[i]n all criminal prosecu- tions, the accused” has “the right to a speedy and public trial, by an impartial jury.” Inherent in that guarantee is an assurance that any guilty verdict will issue only from a unanimous jury. Ramos v. Louisi- ana, 590 U. S. 83, 93. The Fifth Amendment further promises that the government may not deprive individuals of their liberty without “due process of law.” It safeguards for criminal defendants well-established common-law protections, including the “ancient rule” that the govern- ment must prove to a jury every one of its charges beyond a reasonable doubt. Together, these Amendments place the jury at the heart of our criminal justice system and ensure a judge's power to punish is derived wholly from, and remains always controlled by, the jury and its verdict. Blakely v. Washington, 542 U. S. 296, 306. The Court has repeatedly cautioned that trial and sentencing prac- tices must remain within the guardrails provided by these two Amend- ments. Thus in Apprendi v. New Jersey, 530 U. S. 466, the Court held that a novel “sentencing enhancement” was unconstitutional because it violated the rule that only a jury may fnd “facts that increase the pre- scribed range of penalties to which a criminal defendant is exposed.” Id., at 490. This principle applies when a judge seeks to issue a sen- tence that exceeds the maximum penalty authorized by a jury's fndings as well as when a judge seeks to increase a defendant's minimum punishment. See, e. g., Alleyne v. United States, 570 U. S. 99, 111–113. Pp. 829–834. (b) The government concedes what all of this means for Mr. Erlinger. To trigger ACCA's mandatory minimum, the government had to prove, among other things, that his three predicate convictions were “com- mitted on occasions different from one another.” § 924(e)(1). And as Wooden observed, deciding whether those past offenses occurred on three or more different occasions is a fact-laden task. As the govern- ment recognizes, virtually “any fact” that “increase[s] the prescribed range of penalties to which a criminal defendant is exposed” must be resolved by a unanimous jury beyond a reasonable doubt (or freely admitted in a guilty plea). Apprendi, 530 U. S., at 490. Here, the sentencing court made a factual fnding that Mr. Erlinger's offenses occurred on at least three separate occasions. And as in Apprendi and Alleyne, that factual fnding had the effect of increasing both the maxi- mum and minimum sentences Mr. Erlinger faced. Thus, Mr. Erlinger was entitled to have a jury resolve ACCA's occasions inquiry unani- mously and beyond a reasonable doubt. This Court decides no more than that. Pp. 834–835. (c) Court-appointed amicus cannot avoid this conclusion. Pp. 836–848. Cite as: 602 U. S. 821 (2024) 823

(1) Amicus relies on an exception announced in Almendarez-Torres v. United States, 523 U. S. 224, which he argues permits a judge to fnd certain facts related to a defendant's past offenses, including whether he committed them on different occasions. That decision is an outlier. And the Court has described it as “at best an exceptional departure” from historic practice. Apprendi, 530 U. S., at 487. It persists as a “narrow exception” permitting judges to fnd only “the fact of a prior conviction.” Alleyne, 570 U. S., at 111, n. 1. Pp. 836–839. (2) Amicus responds that if Almendarez-Torres permits a judge to fnd the fact of a conviction, that necessarily implies that a judge may also fnd the jurisdiction in which the underlying offense occurred and the date it happened, which is generally enough to resolve the occasions inquiry, making sending it to a jury pointless. This Court disagrees. To answer such questions, a court will sometimes consult the Shepard documents in a case, which include judicial records, plea agreements, and colloquies between a judge and the defendant. See Shepard v. United States, 544 U. S. 13. This Court's cases hold that a sentenc- ing judge may use the information gleaned from Shepard documents for the “limited function” of determining the fact of a prior conviction and the then-existing elements of that offense. “[N]o more is allowed.” Mathis v. United States, 579 U. S. 500, 511. Moreover, often Shepard documents will not contain all the information needed to conduct a sen- sible ACCA occasions inquiry, and they can also be “prone to error.” Mathis, 579 U. S., at 512. Pp. 839–842. (3) Amicus insists this Court's Almendarez-Torres precedents are mistaken, because the Fifth and Sixth Amendments' original meaning and common-law traditions authorize judges at sentencing to fnd all manner of facts about an offender's past crimes. But this Court has been down this road many times before, and to reconsider all those precedents now would require, at the least, convincing proof indeed. See United States v. Gaudin, 515 U. S. 506, 515. Yet amicus offers nothing like that, and the evidence he does offer does more to hurt than help his cause. Amicus points to supplemental information procedures that a few States employed in the early 19th century. But a sentencing procedure followed by a few States hardly represents “convincing” proof that our precedents have mistaken the original meaning of the Fifth and Sixth Amendments. And in upholding one such scheme, the Court stressed that, under the law's terms, even “the fact of former con- viction” had to be “charged” by prosecutors and then “determined by a jury in a proceeding thereby instituted.” Graham v. West Virginia, 224 U. S. 616, 630. Amicus next turns to the Double Jeopardy Clause, which permits a judge to look into a defendant's past conduct to ask whether the government has charged a defendant for the same crime a 824 ERLINGER v. UNITED STATES

second time. While the Double Jeopardy Clause protects a defendant by prohibiting a judge from even empaneling a jury when the defendant has already faced trial on the charged crime, the Fifth and Sixth Amendments' jury trial rights provide a defendant with entirely comple- mentary protections at a different stage of the proceedings by ensuring that, once a jury is lawfully empaneled, the government must prove beyond a reasonable doubt to a unanimous jury the facts necessary to sustain the punishment it seeks. Finally, amicus points to case law and statutes in four other States. But while this evidence may suggest that in a small number of jurisdictions, judges could fnd the existence, num- ber, and dates of a defendant's prior convictions, none of this provides a persuasive basis for revisiting this Court's many precedents forbidding judges from doing more, let alone prove a longstanding tradition. Pp. 843–847. (4) Amicus argues that leaving the occasions inquiry to juries would do more to prejudice than to protect defendants. That concern, like arguments about effciency, cannot alter the demands of the Fifth and Sixth Amendments. Tools such as bifurcation in any event exist to ad- dress the prejudicial effect evidence about a defendant's past crimes can have on a jury. Pp. 847–848.

77 F. 4th 617, vacated and remanded.

Gorsuch, J., delivered the opinion of the Court, in which Roberts, C. J., and Thomas, Sotomayor, Kagan, and Barrett, JJ., joined. Rob- erts, C. J., post, p. 849, and Thomas, J., post, p. 850, fled concurring opin- ions. Kavanaugh, J., fled a dissenting opinion, in which Alito, J., joined, and in which Jackson, J., joined except as to Part III, post, at 851. Jack- son, J., fled a dissenting opinion, post, p. 871.

Jeffrey L. Fisher argued the cause for petitioner. With him on the briefs were Rachel A. Chung, Jessie A. Cook, Heather Welles, and Louis W. Fisher. Deputy Solicitor General Feigin argued the cause for the United States in support of petitioner. With him on the briefs were Solicitor General Prelogar, Acting Assistant At- torney General Argentieri, Yaira Dubin, and Paul T. Crane. D. Nick Harper, by invitation of the Court, 601 U. S. –––, argued the cause and fled a brief as amicus curiae in sup- port of the judgment below. With him on the brief were Cite as: 602 U. S. 821 (2024) 825

Opinion of the Court

Miguel A. Estrada, Jonathan C. Bond, and Robert A. Batista.* Justice Gorsuch delivered the opinion of the Court. This case concerns the Armed Career Criminal Act (ACCA) and the Fifth and Sixth Amendments. ACCA im- poses lengthy mandatory prison terms on certain defend- ants who have previously committed three violent felonies or serious drug offenses on separate occasions. The ques- tion we face is whether a judge may decide that a defendant's past offenses were committed on separate occasions under a preponderance-of-the-evidence standard, or whether the Fifth and Sixth Amendments require a unanimous jury to make that determination beyond a reasonable doubt.

I In 2017, the federal government charged Paul Erlinger under 18 U. S. C. § 922(g) with being a felon unlawfully in possession of a frearm. At the time, a conviction for that offense ordinarily carried a sentence of up to 10 years in prison. § 924(a)(2) (2012 ed.). But the government also charged Mr. Erlinger under ACCA. And under that stat- ute, a defendant found guilty of violating § 922(g) can face even more severe punishment. Specifcally, if the defendant has three prior convictions for “violent felon[ies]” or “serious drug offense[s]” that were “committed on occasions different from one another,” ACCA increases the prison term he faces to a minimum of 15 years and a maximum of life. § 924(e)(1) (2012 ed.).

*Briefs of amici curiae urging reversal were fled for FAMM by Amy Mason Saharia, Mary Price, and Peter Goldberger; for the National Asso- ciation of Criminal Defense Lawyers by Peter B. Siegal, Jeffrey T. Green, and David Oscar Markus; and for the National Association of Federal Defenders by Davina T. Chen, Shelley M. Fite, Michael C. Holley, Andrew L. Adler, Jennifer Niles Coffn, and Judy Madewell. 826 ERLINGER v. UNITED STATES

Mr. Erlinger pleaded guilty to violating § 922(g). But in doing so, he stressed that his previous convictions were more than a decade old and since then he had turned his life around, secured a steady job, built a family, and remained free from drugs. Tr. of Plea and Sentencing Hearing in No. 2:18–cr–00013 (SD Ind., Oct. 24, 2018), ECF Doc. 67, p. 24. In light of these facts, the sentencing judge stated that a 5-year prison term would represent a “fair sentence.” Id., at 30. But the judge also found it more likely than not that Mr. Erlinger's past included three ACCA-qualifying offenses committed on three different occasions. Id., at 17. And based on that fnding, the court concluded, it had no author- ity to issue a 5-year sentence. Id., at 29–30. Instead, ACCA required a sentence of at least 15 years. That was “too high” for Mr. Erlinger's crime, the court emphasized, but it had “no power” to order anything less. Ibid. Shortly after the court issued a 15-year sentence consist- ent with ACCA, the ground shifted. The Seventh Circuit issued decisions indicating that two of the three offenses on which the district court had relied to invoke ACCA did not qualify as “violent felon[ies]” or “serious drug offense[s].” 77 F. 4th 617, 619 (2023) (citing United States v. Glispie, 978 F. 3d 502 (2020), and United States v. De La Torre, 940 F. 3d 938 (2019)). That left Mr. Erlinger with only one qualifying prior conviction, not the three ACCA requires. Recogniz- ing as much, the district court vacated its sentence and set the case for a new sentencing hearing. Order, ECF Doc. 81. There, prosecutors once more pursued a 15-year ACCA sentence. This time, though, they pointed to a new set of prior convictions arising from burglaries Mr. Erlinger com- mitted when he was 18 years old—some 26 years before his § 922(g) charge. Tr. of Video Resentencing Hearing, ECF Doc. 120, at 37–38. As the government told it, within a span of days Mr. Erlinger burglarized a pizza shop, a sporting goods store, and two restaurants. Ibid. Because each of Cite as: 602 U. S. 821 (2024) 827

these burglaries occurred on different occasions, the govern- ment submitted, each could serve as an ACCA predicate and collectively they could support an ACCA sentence. Id., at 39–40. Mr. Erlinger protested. He maintained that his decades- old burglaries had not occurred on four separate occasions but during a single criminal episode, all of which meant he still lacked the three prior offenses ACCA requires. Id., at 20. More than that, Mr. Erlinger argued, the question whether he committed his prior burglaries during a single episode or on distinct occasions required an assessment of the facts surrounding those offenses. Id., at 22–23. And, he contended, the Fifth and Sixth Amendments entitled him to have a jury make that assessment. Id., at 23. The district court rejected Mr. Erlinger's request for a jury. It proceeded to fnd for itself that each of his 26-year- old burglaries occurred on distinct occasions. Id., at 48–49. Armed with this fnding, the court held that Mr. Erlinger had at least three previous ACCA-qualifying offenses and once more issued a 15-year sentence. Again, the court said that it thought the result “unfortunate” and “excessive.” Id., at 50. But, again, it said ACCA left it no choice. Id., at 50–51. On appeal, Mr. Erlinger renewed his argument that the Fifth and Sixth Amendments entitled him to have a jury decide whether his burglaries occurred on different occa- sions. But this time, the government confessed error. The Constitution, it said, “requires a jury” to decide unanimously and beyond a reasonable doubt whether Mr. Erlinger 's ACCA predicates were “ `committed on occasions different from one another.' ” 77 F. 4th, at 619, 620. In explaining its turnabout, the government pointed to this Court's recent decision in Wooden v. United States, 595 U. S. 360 (2022). That decision did not directly address whether a judge may, or a jury must, resolve disputes about 828 ERLINGER v. UNITED STATES

whether multiple crimes occurred on multiple occasions. Id., at 365, n. 3; see id., at 397, n. 7 (Gorsuch, J., concurring in judgment). But, the government acknowledged, Wooden did hold that ACCA's occasions “inquiry” can require an examination of a “range” of facts, including whether the de- fendant's past offenses were “committed close in time,” whether they were committed near to or far from one an- other, and whether the offenses were “similar or inter- twined” in purpose and character. Id., at 369. And given the intensely factual nature of this inquiry and the impact its resolution can have on a defendant's sentence, the govern- ment admitted, a jury must resolve it. That conclusion, the government represented, fows directly from this Court's consistent holdings that the Fifth and Sixth Amendments generally guarantee a defendant the right to have a unani- mous jury fnd beyond a reasonable doubt any fact that in- creases his exposure to punishment. Brief for United States in No. 22–1926 (CA7), pp. 9–11. Despite the government's concession, the Seventh Circuit refused to disturb the district court's sentence. 77 F. 4th, at 621–622. That left Mr. Erlinger to petition this Court for certiorari. The government fled a brief in support of his petition. In it, the government argued that a number of courts of appeals have refused requests for juries in cases like Mr. Erlinger's and that “this Court's intervention is nec- essary to ensure that the circuits correctly recognize defend- ants' constitutional rights in this context.” Brief for United States on Pet. for Cert. 5–6. We agreed to take up Mr. Erlinger 's case to decide whether ACCA's occasions inquiry must be resolved by a jury. 601 U. S. ––– (2023). Because the government now agrees with Mr. Erlinger about the proper resolution of that question, we appointed Nick Harper to defend the judgment below as amicus curiae. 601 U. S. ––– (2023). He has ably discharged his responsibilities. Cite as: 602 U. S. 821 (2024) 829

II A Prominent among the reasons colonists cited in the Decla- ration of Independence for their break with Great Britain was the fact Parliament and the Crown had “depriv[ed] [them] in many cases, of the benefts of Trial by Jury.” ¶20. For centuries, English law had recognized the right to trial by jury. Duncan v. Louisiana, 391 U. S. 145, 151 (1968). Yet, as tensions grew between the British Empire and its American Colonies, imperial authorities responded by strip- ping away that ancient right. By their lights, colonial juries “ `were not to be trusted' ” because they found for defend- ants too often. D. Lovejoy, Rights Imply Equality: The Case Against Admiralty Jurisdiction in America, 1764–1776, 16 Wm. & Mary Q. 459, 468 (1959). To secure more vigorous enforcement of the Stamp Act and other unpopular laws, authorities directed more and more cases to vice-admiralty courts where crown-appointed judges, rather than local ju- ries, decided the defendant's fate. Jones v. United States, 526 U. S. 227, 245–246 (1999). Just as authorities hoped, the tactic proved “ `most effective' ” at securing the verdicts they wished. Parklane Hosiery Co. v. Shore, 439 U. S. 322, 340, n. 3 (1979) (Rehnquist, J., dissenting) (quoting 11 W. Holdsworth, A History of English Law 110 (1966)). After securing their independence, the founding genera- tion sought to ensure what happened before would not hap- pen again. As John Adams put it, the founders saw repre- sentative government and trial by jury as “the heart and lungs” of liberty. Letter from Clarendon to W. Pym (Jan. 27, 1766), in 1 Papers of John Adams 169 (R. Taylor ed. 1977). “[W]ithout them,” he wrote, we “have no other fortifcation . . . against being ridden like horses, feeced like sheep, worked like cattle, and fed and clothed like swine and hounds.” Ibid. Refecting that sentiment, the right to trial 830 ERLINGER v. UNITED STATES

by jury in criminal cases was, on one telling, the only right included in every newly enacted state constitution. A. Al- schuler & A. Deiss, A Brief History of the Criminal Jury in the United States, 61 U. Chi. L. Rev. 867, 870 (1994). Those who drafted our Federal Constitution took just as strong a stand on the jury trial right. As originally pro- posed, the Constitution promised that “[t]he Trial of all Crimes, except in Cases of Impeachment, shall be by Jury.” Art. III, § 2, cl. 3. In the ratifcation debates that followed, some questioned the adequacy of this provision; even with it, they feared, the new federal government might fall prey to the kinds of temptations that led the British to restrict the jury trial right in the colonies. That right, they argued, had to be “guard[ed] with the most jealous circumspection.” A [New Hampshire] Farmer, No. 3, June 6, 1788, quoted in The Complete Bill of Rights 681 (N. Cogan 2d ed. 2015) (in- ternal quotation marks and italics omitted). To address this and other concerns about the new Constitution, James Madi- son agreed to draft a series of amendments we now know as the Bill of Rights. No fewer than three of those ten amend- ments touch on the right to trial by jury, two with implica- tions for criminal cases. Amdts. 5, 6, 7. Madison himself described protections for the jury trial right as among “the most valuable” that appear in “the whole list” of amend- ments he produced. 1 Annals of Cong. 755 (1789). The Sixth Amendment promises that “[i]n all criminal prosecutions, the accused” has “the right to a speedy and public trial, by an impartial jury.” Inhering in that guaran- tee is an assurance that a guilty verdict will issue only from a unanimous jury. Ramos v. Louisiana, 590 U. S. 83, 93 (2020). The Fifth Amendment further promises that the government may not deprive individuals of their liberty without “due process of law.” It is a promise that safe- guards for criminal defendants those procedural protections well established at common law, including the “ancient rule” that the government must prove to a jury every one of its Cite as: 602 U. S. 821 (2024) 831

charges beyond a reasonable doubt. United States v. Hay- mond, 588 U. S. 634, 641 (2019) (plurality opinion); see Ap- prendi v. New Jersey, 530 U. S. 466, 477–478 (2000); United States v. Gaudin, 515 U. S. 506, 510 (1995); Sullivan v. Loui- siana, 508 U. S. 275, 277–278 (1993). The Fifth and Sixth Amendments placed the jury at the heart of our criminal justice system. From the start, those provisions were understood to require the government to in- clude in its criminal charges “ `all the facts and circumstances which constitute the offence.' ” Apprendi, 530 U. S., at 478 (quoting J. Archbold, Pleading and Evidence in Criminal Cases 44 (15th ed. 1862)). Should an “indictment or `accusa- tion . . . lack any particular fact which the laws ma[d]e es- sential to the punishment,' it was treated as `no accusation' at all.” Haymond, 588 U. S., at 642 (quoting 1 J. Bishop, Criminal Procedure § 87, p. 55 (2d ed. 1872) (some alterations omitted)). And at all times the “ `truth of every accusa- tion' ” against a defendant had to be “ `confrmed by the unan- imous suffrage of twelve of [his] equals and neighbours.' ” Apprendi, 530 U. S., at 477 (quoting 4 W. Blackstone, Com- mentaries on the Laws of England 343 (1769); emphasis deleted). Equally, the Fifth and Sixth Amendments sought to en- sure that a judge's power to punish would “deriv[e] wholly” from, and remain always “control[led]” by, the jury and its verdict. Blakely v. Washington, 542 U. S. 296, 306 (2004). Ordinarily, common-law crimes carried “specifc sanctions, and `[o]nce the facts of the offense were determined by the jury, the judge was meant simply to impose the prescribed sentence.' ” Haymond, 588 U. S., at 642 (quoting Alleyne v. United States, 570 U. S. 99, 108 (2013) (plurality opinion)). Even when it came to early American statutes that gave judges a measure of discretion in selecting among possible sentences, “the ranges themselves were linked to particular facts” found by the jury. Alleyne, 570 U. S., at 109 (collect- ing examples). All of which guaranteed that a judge could 832 ERLINGER v. UNITED STATES

not “ `swell the penalty above what the law . . . provided for the acts' ” found by a jury of the defendant's peers. Hay- mond, 588 U. S., at 642 (quoting Apprendi, 530 U. S., at 519 (Thomas, J., concurring)). These principles represent not “procedural formalit[ies]” but “fundamental reservation[s] of power” to the American people. Blakely, 542 U. S., at 305–306. By requiring the Executive Branch to prove its charges to a unanimous jury beyond a reasonable doubt, the Fifth and Sixth Amendments seek to mitigate the risk of prosecutorial overreach and mis- conduct, including the pursuit of “pretended offenses” and “arbitrary convictions. ” The Federalist No. 83, p. 499 (C. Rossiter ed. 1961); accord, Federal Farmer, Letter XV (Jan. 18, 1788), reprinted in 2 The Complete Anti-Federalist 320 (H. Storing ed. 1981). By requiring a unanimous jury to fnd every fact essential to an offender's punishment, those amendments similarly seek to constrain the Judicial Branch, ensuring that the punishments courts issue are not the result of a judicial “inquisition” but are premised on laws adopted by the people's elected representatives and facts found by members of the community. Blakely, 542 U. S., at 307; Hay- mond, 588 U. S., at 640–641. Both of these checks on gov- ernmental power, the framers appreciated, were “anchor[s]” essential to prevent a slide back toward regimes like the vice-admiralty courts they so despised. Letter from T. Jef- ferson to T. Paine (July 11, 1789), reprinted in 15 Papers of Thomas Jefferson 266, 269 (J. Boyd ed. 1958). With the passage of time, and accelerating in earnest in the 20th century, various governments in this country sought to experiment with new trial and sentencing practices. See Mistretta v. United States, 488 U. S. 361, 363–367 (1989); Wil- liams v. New York, 337 U. S. 241, 247–248 (1949); see also P. Tappan, Sentencing Under the Model Penal Code, 23 Law & Contemp. Prob. 528, 529–532 (1958). But in case after case, this Court has cautioned that, while some experi- ments may be tolerable, all must remain within the Fifth and Sixth Amendments' guardrails. Cite as: 602 U. S. 821 (2024) 833

So, for example, in Apprendi this Court faced a case in- volving a New Jersey offense that ordinarily carried a maxi- mum sentence of 10 years in prison. 530 U. S., at 468. The State, however, had adopted a novel “sentencing enhance- ment” that purported to allow a judge to impose an even longer term of imprisonment after fnding, by a preponder- ance of the evidence, that the offender's crime was motivated by racial bias. Id., at 468–471. Relying on that statutory authority, the sentencing judge ordered the defendant to serve 12 years in prison. Id., at 471. This Court found the sentence unconstitutional, and did so for a by-now familiar reason: Only a jury may fnd “ `facts that increase the pre- scribed range of penalties to which a criminal defendant is exposed.' ” Id., at 490. It is a principle we have since reiterated in response to a variety of other recent sentencing innovations. See Hay- mond, 588 U. S., at 644 (collecting cases). And it is a princi- ple, we have observed, that does not just apply when a judge seeks to issue a sentence that exceeds the maximum penalty authorized by a jury's fndings (or a guilty plea). It is a principle that also applies when a judge seeks to increase a defendant's minimum punishment. Alleyne illustrates the point. There, we confronted a case in which a jury had con- victed the defendant of a crime that usually carried a sen- tence of between fve years and life in prison. 570 U. S., at 103–104. But a separate statutory “sentencing enhance- ment” ostensibly allowed the judge to transform that 5-year minimum sentence into a 7-year minimum sentence if he found a certain additional fact by a preponderance of the evidence. Ibid. That innovation, too, the Court held, im- properly invaded the jury's province because “[a] fact that increases” a defendant's exposure to punishment, whether by triggering a higher maximum or minimum sentence, must “be submitted to a jury” and found unanimously and beyond a reasonable doubt. Id., at 111–113. The principles Apprendi and Alleyne discussed are so frmly entrenched that we have now overruled several deci- 834 ERLINGER v. UNITED STATES

sions inconsistent with them. See, e. g., Hurst v. Florida, 577 U. S. 92, 101–102 (2016) (overruling Hildwin v. Florida, 490 U. S. 638 (1989) (per curiam), and Spaziano v. Florida, 468 U. S. 447 (1984)); Alleyne, 570 U. S., at 107 (overruling Harris v. United States, 536 U. S. 545 (2002)); Ring v. Ari- zona, 536 U. S. 584, 609 (2002) (overruling Walton v. Ari- zona, 497 U. S. 639 (1990)). B Commendably, the government concedes before us, as it did before the court of appeals, what all this means for Mr. Erlinger 's case and others like it. Under § 922(g), Mr. Erlinger faced between 0 and 10 years in prison. § 924(a)(2) (2012 ed.). To trigger ACCA and expose him to longer prison terms, the government had to prove that his past included three convictions for “violent felon[ies]” or “se- rious drug offense[s]” that were “committed on occasions dif- ferent from one another.” § 924(e)(1). And under Wooden, deciding whether those past offenses occurred on three or more different occasions is a fact-laden task. Were the crimes “committed close in time”? 595 U. S., at 369. How about the “[p]roximity” of their “location[s]”? Ibid. Were the offenses “similar or intertwined” in purpose and charac- ter? Ibid. All these questions, Wooden observed, “may be relevant” to determining whether the offenses were com- mitted on one occasion or separate ones—and all require facts to be found before ACCA's more punitive mandatory minimum sentence may be lawfully deployed. Ibid. As the government recognizes, there is no doubt what the Constitution requires in these circumstances: Virtually “any fact” that “ `increase[s] the prescribed range of penalties to which a criminal defendant is exposed' ” must be resolved by a unanimous jury beyond a reasonable doubt (or freely ad- mitted in a guilty plea). Apprendi, 530 U. S., at 490; see Brief for United States 9. Judges may not assume the jury's factfnding function for themselves, let alone purport to per- form it using a mere preponderance-of-the-evidence stand- ard. To hold otherwise might not portend a revival of the Cite as: 602 U. S. 821 (2024) 835

vice-admiralty courts the framers so feared. See Part II– A, supra. But all the same, it would intrude on a power the Fifth and Sixth Amendments reserve to the American people. Really, this case is as nearly on all fours with Apprendi and Alleyne as any we might imagine. In Apprendi, a judge relied on his own factual fndings under a preponderance-of- the-evidence standard to increase the defendant's maximum sentence from 10 to 20 years. 530 U. S., at 469. In Alleyne, a judge proceeded the same way to increase the defendant's minimum sentence from fve to seven years. 570 U. S., at 104. Here, the sentencing court's factual fnding that Mr. Erlinger's offenses occurred on at least three separate occasions had the effect of increasing both the maximum and minimum sentences he faced. Rather than a maximum sen- tence of 10 years in prison, the judge's fnding left Mr. Er- linger exposed to life in prison. Rather than a minimum penalty of no prison time, the judge's fnding meant Mr. Er- linger had to serve at least 15 years. While recognizing Mr. Erlinger was entitled to have a jury resolve ACCA's occasions inquiry unanimously and beyond a reasonable doubt, we decide no more than that. For pur- poses of the proceedings before us, the parties take as given that Mr. Erlinger committed four burglaries and that each qualifes as a “violent offense” under ACCA. But they dis- agree vigorously about whether those burglaries took place on at least three different occasions (so that ACCA's en- hanced sentences would apply) or during a single criminal episode (so that they would not). Presented with evidence about the times, locations, purpose, and character of those crimes, a jury might have concluded that some or all oc- curred on different occasions. Or it might not have done so. All we can say for certain is that the sentencing court erred in taking that decision from a jury of Mr. Erlinger's peers.1 1 Justice Jackson pursues an argument neither the government nor amicus nor the principal dissent attempts. She says Apprendi v. New Jersey, 530 U. S. 466 (2000), “was wrongly decided,” and all but calls on 836 ERLINGER v. UNITED STATES

III A The Court-appointed amicus resists the conclusion we reach. Unlike Justice Jackson, see n. 1, supra, however, amicus does not dispute the Constitution's time-honored guarantee that a unanimous jury ordinarily must fnd beyond a reasonable doubt any fact that increases a defendant's ex- posure to punishment. See Brief for Court-Appointed Ami-

the Court to overturn our many precedents applying it, post, at 872, and n. 1, 880–889. But rather than meaningfully engage with the Constitu- tion, its original meaning and history, or our precedents, Justice Jackson would abandon “constitutional theory” and appeal to a different authority. Post, at 897. “In my view,” post, at 890, Justice Jackson contends, ju- ries cannot “deal with the fne-grained, nuanced determinations . . . that are necessary to fairly adjudicate factual questions like the one that ACCA's occasions inquiry raises,” post, at 893. But the Constitution does not take such a dim view about the capacity of jurors or the rigors of trial. Surely, too, juries are no less capable than judges to decide whether three past events happened on three separate occasions. Day in and day out, using everyday trial procedures, juries decide exponentially more complex questions than that. Nor, of course, does Apprendi prohibit legislatures from enacting reforms authorizing judges to lower sentences based on their own factfnding. See, e. g., First Step Act of 2018, Pub. L. 115–391, 132 Stat. 5194; 18 U. S. C. § 3553(f) (doing just that). The only thing judges may not do consistent with Apprendi is increase a defendant's ex- posure to punishment based on their own factfnding. All of which leaves Justice Jackson with her real complaint: In her view, it is “wildly ineff- cient” to require the government to call witnesses and present evidence— which is to say prove its case—when a judge might more easily enhance a defendant's exposure to punishment by consulting “sometimes-decades- old,” post, at 893, and error-prone court records, infra, at 840–842. But does Justice Jackson really think it too much to ask the government to prove its case (as it concedes it must) with reliable evidence before seeking enhanced punishments under a statute like ACCA when the “practical realit[y]” for defendants like Mr. Erlinger is exposure to an additional decade (or more) in prison? Post, at 897. Justice Jackson may view juries as “roadblocks” to higher punishments. Post, at 889. But “[t]he bottom line is this”: the people ratifed the Fifth and Sixth Amendments, not any of our personal views. Post, at 897. Cite as: 602 U. S. 821 (2024) 837

cus Curiae 1. In defending the decision below, amicus relies instead on an exception to that rule this Court announced in Almendarez-Torres v. United States, 523 U. S. 224 (1998). On amicus's telling, that exception permits a judge to fnd perhaps any fact related to a defendant's past offenses, including whether he committed them on different occasions. Brief for Court-Appointed Amicus Curiae 10. The principal dissent presses the same line of argument. Post, at 855 (opinion of Kavanaugh, J.). We disagree. In Almendarez-Torres, the Court consid- ered sentencing laws applicable to aliens who returned to the United States after a previous removal. The default sentencing range was up to two years of imprisonment. 8 U. S. C. § 1326(a) (1994 ed.). But a fnding that the govern- ment previously removed the alien after a “conviction for commission of an aggravated felony” triggered a new maxi- mum penalty of up to 20 years in prison. § 1326(b) (1994 ed.). In Almendarez-Torres, the Court permitted a judge to undertake the job of fnding the fact of a prior convic- tion—and that job alone. 523 U. S., at 246–247. Almost immediately, too, the decision came under scrutiny. Jones, 526 U. S., at 249, n. 10. The Court has since de- scribed Almendarez-Torres as “at best an exceptional depar- ture” from “historic practice.” Apprendi, 530 U. S., at 487. That decision, we have said, parted ways from the “uniform course of decision during the entire history of our jurispru- dence.” Id., at 490. It was “arguabl[y] . . . incorrec[t].” Id., at 489. And it amounted to an “unusual . . . exception to the Sixth Amendment rule in criminal cases that `any fact that increases the penalty for a crime' must be proved to a jury.” Pereida v. Wilkinson, 592 U. S. 224, 238 (2021) (quot- ing Apprendi, 530 U. S., at 490). In separate opinions, a number of Justices have criticized Almendarez-Tor res further yet, and Justice Thomas, whose vote was essential to the majority in that case, has called for it to be overruled. See, e. g., Mathis v. United 838 ERLINGER v. UNITED STATES

States, 579 U. S. 500, 522 (2016) (Thomas, J., concurring); Des- camps v. United States, 570 U. S. 254, 280 (2013) (Thomas, J., concurring in judgment); Shepard v. United States, 544 U. S. 13, 27 (2005) (Thomas, J., concurring in part and concur- ring in judgment); see also Jones, 526 U. S., at 252–253 (Stevens, J., concurring); Monge v. California, 524 U. S. 721, 741 (1998) (Scalia, J., joined by Souter and Ginsburg, JJ., dissenting). Still, no one in this case has asked us to revisit Almendarez- Torres. Nor is there need to do so today. In the years since that decision, this Court has expressly delimited its reach. It persists as a “narrow exception” permitting judges to fnd only “the fact of a prior conviction.” Alleyne, 570 U. S., at 111, n. 1. Under that exception, a judge may “do no more, consistent with the Sixth Amendment, than de- termine what crime, with what elements, the defendant was convicted of.” Mathis, 579 U. S., at 511–512. We have reit- erated this limit on the scope of Almendarez-Torres “over and over,” to the point of “downright tedium.” 579 U. S., at 510, 519.2 And so understood, Almendarez-Torres does nothing to save the sentence in this case. To determine whether Mr. Erlinger's prior convictions triggered ACCA's enhanced penalties, the district court had to do more than identify his previous convictions and the legal elements re- quired to sustain them. It had to fnd that those offenses

2 See Pereida v. Wilkinson, 592 U. S. 224, 238 (2021); United States v. Haymond, 588 U. S. 634, 644, n. 3 (2019) (plurality opinion); Descamps v. United States, 570 U. S. 254, 269 (2013); Cunningham v. California, 549 U. S. 270, 282 (2007); Shepard v. United States, 544 U. S. 13, 24 (2005) (plurality opinion); United States v. Booker, 543 U. S. 220, 244 (2005); Blakely v. Washington, 542 U. S. 296, 301 (2004); Apprendi v. New Jersey, 530 U. S. 466, 490 (2000). The principal dissent complains that some of these decisions “did not involve recidivism” questions. Post, at 857 (opinion of Kavanaugh, J.). But surely our consistent explanations of Almendarez-Torres's reach cannot be so casually dismissed. Perhaps es- pecially when those explanations were repeated in multiple cases that did “involve recidivism,” such as Shepard, Descamps, and Mathis. Cite as: 602 U. S. 821 (2024) 839

occurred on at least three separate occasions. And, in doing so, the court did more than Almendarez-Torres allows.

B Perhaps, amicus replies, but perhaps not. If Almendarez- Torres permits a judge to fnd “what crime, with what ele- ments, the defendant was convicted of,” Mathis, 579 U. S., at 511–512, amicus reasons, that necessarily implies a judge may also fnd the jurisdiction in which the underlying offense occurred and the date it happened. Brief for Court- Appointed Amicus Curiae 29. And, amicus continues, in at least some (but admittedly not all) cases, knowing those facts will make the occasions inquiry so “ `straightforward' ” that sending it to a jury would be pointlessly ineffcient. Id., at 39–40. Again, we disagree. To conduct the narrow inquiry Almendarez-Torres authorizes, a court may need to know the jurisdiction in which the defendant's crime occurred and its date in order to ascertain what legal elements the govern- ment had to prove to secure a conviction in that place at that time. And to answer those questions, a sentencing court may sometimes consult “a restricted set of materials,” often called Shepard documents, that include judicial records, plea agreements, and colloquies between a judge and the defend- ant. Descamps, 570 U. S., at 262; see Shepard, 544 U. S., at 20–21, 26; Taylor v. United States, 495 U. S. 575, 602 (1990). None of that, however, means that a court may use Shep- ard documents or any other materials for any other purpose. To ensure compliance with the Fifth and Sixth Amendments, a sentencing judge may use the information he gleans from Shepard documents for the “limited function” of determining the fact of a prior conviction and the then-existing elements of that offense. Descamps, 570 U. S., at 260. “[N]o more” is allowed. Mathis, 579 U. S., at 511. In particular, a judge may not use information in Shepard documents to decide “what the defendant . . . actually d[id],” or the “means” or 840 ERLINGER v. UNITED STATES

“manner” in which he committed his offense in order to in- crease the punishment to which he might be exposed. 579 U. S., at 504, 510–511; see Descamps, 570 U. S., at 269. To sanction that practice would be to allow a sentencing court to do exactly what the Fifth and Sixth Amendments forbid. Ibid. The sentencing court in this case disregarded these con- straints. To determine what legal elements attached to Mr. Erlinger's decades-old offenses, the court might have needed to consult Shepard documents to ascertain the juris- diction in which they occurred and the date on which they happened. But the court had no need or authority “to go any further,” Mathis, 579 U. S., at 511, and assume for itself the responsibility of deciding whether Mr. Erlinger's past offenses differed enough in time, location, character, and pur- pose to have transpired on different occasions. Let alone undertake that inquiry all with an eye toward increasing his punishment. The Fifth and Sixth Amendments “contem- plat[e] that a jury—not a sentencing court—will fnd such facts, unanimously and beyond a reasonable doubt.” Des- camps, 570 U. S., at 269.3 Other considerations fortify our conclusion. Often, as amicus concedes, Shepard documents will not contain all the information needed to conduct a sensible ACCA occasions inquiry, such as the exact times and locations of the defend- ant's past crimes. Brief for Court-Appointed Amicus Cu-

3 The principal dissent dismisses our precedents restricting the materi- als and facts judges can consider on the ground that those decisions did not “purpor[t] to articulate any Sixth Amendment holding.” Post, at 858 (opinion of Kavanaugh, J.). But, as the government recognizes, this Court did address the Sixth Amendment and “meant what it said.” See Reply Brief for United States 7. As the Court has said before and we hold again today: “[A] judge cannot go beyond identifying the crime of conviction to explore the manner in which the defendant committed that offense. . . . He can do no more, consistent with the Sixth Amendment, than determine what crime, with what elements, the defendant was con- victed of.” Mathis v. United States, 579 U. S. 500, 511–512 (2016). Cite as: 602 U. S. 821 (2024) 841

riae 40. Even when Shepard documents do contain that kind of granular information, more still may be required. After all, this Court has held that no particular lapse of time or distance between offenses automatically separates a sin- gle occasion from distinct ones. Wooden, 595 U. S., at 369– 370. Often, a qualitative assessment about “the character and relationship” of the offenses may be required. Id., at 369. So may an inquiry into whether the crimes shared “a common scheme or purpose.” Ibid. Not only are Shepard documents of limited utility, they can be “prone to error.” Mathis, 579 U. S., at 512; see also Brief for National Association of Federal Defenders as Ami- cus Curiae 8–15 (NAFD Brief) (recounting examples of ma- terial errors); post, at 895 (opinion of Jackson, J.) (acknowl- edging records are “imperfect” and may contain “material gaps”). The risk of error may be especially grave when it comes to facts recounted in Shepard documents on which adversarial testing was “unnecessary” in the prior proceed- ing. Mathis, 579 U. S., at 512. As we have recognized, “[a]t trial, and still more at plea hearings, a defendant may have no incentive to contest what does not matter” to his convic- tion at the time. Ibid. He may even “ `have good reason not to' ” haggle over seemingly immaterial errors in his judi- cial records. Ibid. (quoting Descamps, 570 U. S., at 270). Those realities counsel caution in the use of Shepard docu- ments. At the time of his prior conviction, a defendant might not have cared if a judicial record contained a mistake about, say, the time or location of his offense. Back then, fne details like those might not have mattered a bit to his guilt or innocence. Contesting them needlessly, too, might have risked squandering the patience and good will of a jury or the judge responsible for pronouncing a sentence. Yet, years later and faced with an ACCA charge, those kinds of details can carry with them life-altering consequences. For Mr. Erlinger, they may mean perhaps 10 more years in prison. As a matter of fair notice alone, old recorded de- 842 ERLINGER v. UNITED STATES

tails, prone to error, sometimes untested, often inessential, and the consequences of which a defendant may not have appreciated at the time, “should not come back to haunt [him] many years down the road by triggering a lengthy manda- tory sentence.” 579 U. S., at 512; see also Jones, 526 U. S., at 249. We do not question amicus's assessment that in many cases the occasions inquiry will be “ `straightforward.' ” Brief for Court-Appointed Amicus Curiae 39. Often, a de- fendant's past offenses will be different enough and sepa- rated by enough time and space that there is little question he committed them on separate occasions. But none of that means a judge rather than a jury should make the call. There is no effciency exception to the Fifth and Sixth Amendments. In a free society respectful of the individual, a criminal defendant enjoys the right to hold the government to the burden of proving its case beyond a reasonable doubt to a unanimous jury of his peers “ `regardless of how over- whelmin[g]' ” the evidence may seem to a judge. Rose v. Clark, 478 U. S. 570, 578 (1986).4

4 The principal dissent stresses that, despite our repeated admonitions about the limitations associated with the use of Shepard materials, “all 12 Courts of Appeals” have permitted judges to use them to resolve the occa- sions inquiry. Post, at 856 (opinion of Kavanaugh, J.). But most of the decisions the dissent cites issued before Wooden's clarifcation about the nature of the occasions inquiry, a clarifcation even the government now recognizes means a jury is required. And to the extent some lower courts have continued to allow judges to resolve occasions inquiries even after Wooden v. United States, 595 U. S. 360 (2022), it seems confusion about Almendarez-Torres has played a role in their misapprehension about what the Constitution requires. See, e. g., United States v. Stowell, 82 F. 4th 607, 611 (CA8 2023) (en banc) (Erickson, J., dissenting) (criticizing the ma- jority for treating as “a foregone conclusion” that a judge may continue to fnd facts not proved to the jury to decide the occasions inquiry); United States v. Brown, 77 F. 4th 301, 301–302 (CA4 2023) (statement of Heytens, J.) (noting the “uncertain scope of Almendarez-Torres”); id., at 302 (Nie- meyer, J., concurring in part) (conceding “different approach[es]” exist “as to the scope of Almendarez-Torre[s]”); id., at 303 (Wynn, J., dissenting Cite as: 602 U. S. 821 (2024) 843

C Amicus next asks us to turn from doctrine to history. While our precedents have consistently read Almendarez- Torres as permitting a judge to fnd only the fact of a prior conviction and the elements required to sustain it, he insists our precedents are mistaken. Properly understood, he sub- mits, the Fifth and Sixth Amendments' original meaning and “deep” common-law traditions authorize judges at sentenc- ing to fnd all manner of facts about an offender's past crimes. Brief for Court-Appointed Amicus Curiae 7. We have been down this road before. In Apprendi, the Court carefully studied the original meaning of the Fifth and Sixth Amendments and found that almost “any fact that in- creases the penalty for a crime beyond the prescribed statu- tory maximum” was understood at the time of the Nation's founding to be a fact a jury must fnd. 530 U. S., at 490; see also id., at 499–512 (Thomas, J., concurring). In the years since, we have come to the same conclusion in one decision after another. See, e. g., Mathis, 579 U. S., at 510; supra, at 833–834 (collecting cases). To reconsider all those precedents now would require, at the least, proof “convincing indeed.” Gaudin, 515 U. S., at 515. Yet amicus offers nothing like that. He points to the “supplemental-information” procedures a few States em- ployed “in the early 19th century. ” Brief for Court- Appointed Amicus Curiae 15. Those procedures allowed prosecutors seeking enhanced penalties premised on a de- fendant's past convictions to charge them in “a supplemental, post-conviction information,” rather than in “a pre-conviction indictment.” Ibid. And, amicus stresses, this Court up- held one State's supplemental-information scheme in Gra- ham v. West Virginia, 224 U. S. 616 (1912). The principal

from denial of rehearing en banc) (questioning “the continuing viability of Almendarez-Torres”); see also United States v. Dudley, 5 F. 4th 1249, 1278 (CA11 2021) (Newsom, J., concurring in part and dissenting in part). 844 ERLINGER v. UNITED STATES

dissent echoes the same points. Post, at 864–867 (opinion of Kavanaugh, J.). But, if anything, the evidence amicus cites does more to hurt than help his cause. For one thing, a sentencing pro- cedure followed by a few States hardly represents “con- vincing” proof that our precedents have mistaken the origi- nal meaning of the Fifth and Sixth Amendments. Gaudin, 515 U. S., at 515; see id., at 519. For another, when this Court upheld one of these supplemental-information schemes in Graham, it stressed that, under the law's terms, even “the fact of former conviction” had to be “charged” by prose- cutors and then “determined by a jury in a proceeding thereby instituted.” 224 U. S., at 630. Put those points to- gether and amicus's evidence provides perhaps more reason to question Almendarez-Torres's narrow exception than to expand it.5 Reaching for another tradition, amicus (but not the princi- pal dissent) turns to the Constitution's Double Jeopardy Clause. That Clause, amicus observes, permits a judge to ask whether the government has charged a defendant for the same crime a second time. From this, he suggests, it must be that a judge can also look into the defendant's past con- duct to increase his sentence. Brief for Court-Appointed

5 Amicus emphasizes that supplemental-information procedures became more “widely adopted” in the 1920s after Graham. Brief for Court- Appointed Amicus Curiae 16. But, since Apprendi, this Court has in- sisted on “remain[ing] true” to the Fifth and Sixth Amendment's original meaning and protecting the rights they secure against “ `erosion.' ” 530 U. S., at 483. Accordingly, relatively modern innovations move us little, for they might just as well represent departures from the Constitution's historic protections as evidence of them. Ibid.; accord, Gaudin, 515 U. S., at 518–519. Nor, for that matter, does looking to more modern times do much to help amicus. He does not dispute that, even as late as 1965, juries usually still had to fnd facts about prior convictions, whether under supplemental-information regimes or more traditional ones. See Ap- prendi, 530 U. S., at 489, n. 15; Almendarez-Torres v. United States, 523 U. S. 224, 261 (1998) (Scalia, J., dissenting). Cite as: 602 U. S. 821 (2024) 845

Amicus Curiae 25–26. But that, too, does not follow. The Double Jeopardy Clause protects a defendant by prohibiting a judge from even empaneling a jury when the defendant has already faced trial on the charged crime. See, e. g., Green v. United States, 355 U. S. 184, 188 (1957). The Fifth and Sixth Amendments' jury trial rights provide a defendant with entirely complementary protections at a different stage of the proceedings by ensuring that, once a jury is lawfully empaneled, the government must prove beyond a reasonable doubt to a unanimous jury the facts necessary to sustain the punishment it seeks. Finally, amicus (rejoined now by the principal dissent, see post, at 863–864 (opinion of Kavanaugh, J.)) directs us to case law and statutes in four other States—South Carolina, Louisiana, Alabama, and Kansas. After the Constitution's adoption, amicus suggests, each of these States left “a wide range of recidivism-related issues” for judges, rather than juries, to resolve. Brief for Court-Appointed Amicus Curiae 18. But what does this prove? Here again, amicus points to procedures in less than a handful of States. That is not the kind of “uniform postratifcation practice” that can some- times “shed light upon the meaning” of the Constitution. Gaudin, 515 U. S., at 519. Nor, again, do these practices prove much even taken on their own terms. Some of these States permitted a judge to make “sequencing” determina- tions—deciding, for example, whether the present offense was the defendant's “ `second or subsequent' ” offense. Brief for Court-Appointed Amicus Curiae 14, and n. 1, 18–19.6 Some allowed a judge to fnd whether the defendant had suc- cessfully overturned a prior conviction on appeal or secured

6 See H. Toulmin, Digest of the Laws of the State of Alabama 209 (1823); Ala. Penal Code § 73 (1866); 1859 Kan. Sess. Laws pp. 283–284; 1868 Kan. Sess. Laws pp. 380–381; 6 D. McCord, Statutes at Large of South Carolina 413 (1839); State v. Smith, 8 Rich. 460 (SC 1832); U. Phillips, Revised Stat- utes of Louisiana 155 (1856); 1870 La. Acts p. 206, § 4. 846 ERLINGER v. UNITED STATES

a pardon. Id., at 19–20, and n. 4.7 Some authorized a judge to fnd whether a defendant's current offense and past crime occurred within a specifed period of time. Id., at 19.8 All told, amicus's evidence may suggest that in a small number of jurisdictions judges could fnd the existence, number, and dates of a defendant's prior convictions. But none of this provides a persuasive basis for revisiting our many prece- dents prohibiting judges from doing more. Let alone prove “ `a longstanding tradition' ” in this Nation allowing a judge to fnd any fact regarding a defendant's “recidivis[m].” Post, at 855 (opinion of Kavanaugh, J.). For that matter, it is not clear whether these four States always allowed judges to fnd even the fact of a defendant's prior conviction. Take South Carolina. In State v. Smith, 8 Rich. 460 (SC 1832), the court seemed to hold that the gov- ernment did not have to allege in its indictment that the defendant had been previously convicted “for the crime of horse stealing” because that was a question for “the Court,” id., at 460–461. But, as another State's supreme court rec- ognized, in so holding Smith may have gone “too far.” State v. Burgett, 22 Ark. 323, 324 (1860). It is unclear, too, whether Smith even accurately refected South Carolina's customary practice. Apprendi, 530 U. S., at 509, n. 5 (Thomas, J., concurring). Similar problems attend amicus's reliance on historic Louisiana practices. In State v. Hudson, 32 La. Ann. 1052 (1880), the Louisiana Supreme Court held that a jury did not “ha[ve] to pass” upon the existence of “previous convictions.” Id., at 1053. But Louisiana “over- ruled” Hudson three decades later, calling it out of step with “the common law” and “other jurisdictions.” State v. Com- pagno, 125 La. 669, 671–672, 51 So. 681, 682 (1910). Later still, amicus contends, Louisiana revived Hudson in State v. Guidry, 169 La. 215, 222, 124 So. 832, 835 (1929). See Brief for Court-Appointed Amicus Curiae 14. But however that 7 See 1840 Ala. Sess. Laws 153; 1868 Kan. Sess. Laws pp. 380–381; State v. Hudson, 32 La. Ann. 1052, 1053 (1880). 8 Toulmin, Digest of the Laws of the State of Alabama, at 377. Cite as: 602 U. S. 821 (2024) 847

may be, the historical practices in the four States amicus highlights do not appear to have been nearly as uniform or expansive as he supposes. D Finally, amicus asks us to consider some practical prob- lems. Most especially, he argues that leaving the occasions inquiry to juries would do more to prejudice than protect defendants. Id., at 41–47. It would because requiring prosecutors to prove that the defendant's prior crimes took place on distinct occasions would enable them “to regale ju- ries with the details” of the defendant's past misconduct. Id., at 42; see also post, at 867–870 (opinion of Kavanaugh, J.); post, at 889–893 (opinion of Jackson, J.). But just as arguments from effciency cannot alter the de- mands of the Fifth and Sixth Amendments, neither may that practical concern, “of which earlier courts were well aware.” Apprendi, 530 U. S., at 521 (Thomas, J., concurring). It is hard not to wonder, too: Are we really to suppose that the amici supporting Mr. Erlinger in this Court, including the National Association of Criminal Defense Lawyers and the National Association of Federal Defenders, have been “some- how duped” into advocating for a rule that would be “unfair to criminal defendants”? Blakely, 542 U. S., at 312; see also NAFD Brief 25; Brief for National Association of Criminal Defense Lawyers as Amicus Curiae 16–17. Of course not. As these groups stress, and the govern- ment agrees, traditional tools exist to address the prejudicial effect evidence about a defendant's past crimes can have on a jury. Most obviously, a court can bifurcate the proceed- ings. In that “common,” Apprendi, 530 U. S., at 521, n. 10 (Thomas, J., concurring), and often “fairest” practice, Spen- cer v. Texas, 385 U. S. 554, 567 (1967), a jury is frst tasked with assessing whether the government has proved the ele- ments of the § 922(g) felon-in-possession charge. Then, and only if it fnds the defendant guilty, the jury turns to consider evidence regarding whether the defendant's prior offenses occurred on different occasions for purposes of applying 848 ERLINGER v. UNITED STATES

ACCA's mandatory minimum sentence under § 924(e). Jus- tice Jackson expresses concern about the “burdens” pro- ceeding this way might impose. See post, at 892. But by sequencing and separating the jury's determinations, a court decreases the likelihood that a jury will be “ `overper- suade[d]' ” by the defendant's prior criminal conduct. Post, at 893 (Jackson, J., dissenting) (quoting Michelson v. United States, 335 U. S. 469, 476 (1948)). On this, all sides agree. The government reports that it “generally agree[s] to bifurcation” in ACCA cases like this one and that it has not “been able to anticipate as to why [it] wouldn't agree to bifurcation.” Tr. of Oral Arg. 58–59. Similarly, the National Association of Federal Defenders reports that, “[t]o [its] knowledge, every jury trial but one has been bifurcated when the jury was permitted to decide both the § 922(g) unlawful-frearm-possession question and the § 924(e) `occasions' question.” NAFD Brief 26.9 States that “have not already done so can likewise adjust to any state-law implications of [our] straightforward application of Apprendi to . . . ACCA's different-occasions requirement.” Reply Brief 21; see also Spencer, 385 U. S., at 586 (Warren, C. J., concurring and dissenting) (observing that, whether or not required to do so, at the time most States had adopted “procedures which postpone the introduction of prior convic- tions until after the jury has found the defendant guilty of the crime currently charged”).10

9 Of the trials where proceedings were bifurcated, the National Associa- tion of Federal Defenders observes, some “have resulted in acquittal or dismissal” at the frst stage where prejudicial past conduct was omitted; another has resulted in acquittal at the second stage after the jury found the government had not carried its burden of showing the defendant's prior offenses were committed on separate occasions; three others “have resulted in guilty verdicts” at both stages. NAFD Brief 21–22. 10 Once more parting ways with the parties and defense bar amici, Jus- tice Jackson asserts that Apprendi's rule confning judges at sentencing to the facts found by the jury hurts defendants and makes “racial dispari- ties” in our criminal justice system “worse.” Post, at 886–887. As we Cite as: 602 U. S. 821 (2024) 849

Roberts, C. J., concurring

* The jury trial may have “never been efficient. ” Ap- prendi, 530 U. S., at 498 (Scalia, J., concurring). It may re- quire assembling a group of the defendant's peers to resolve unanimously even seemingly straightforward factual ques- tions under a daunting reasonable doubt standard. Avoid- ing the prejudice associated with the introduction of evi- dence of past crimes may require careful attention, too. But the right to a jury trial “has always been” an important part of what keeps this Nation “free.” Ibid. Because the Fifth and Sixth Amendments do not tolerate the denial of that right in this case, the judgment of the Court of Appeals for the Seventh Circuit is vacated, and the matter is remanded for further proceedings consistent with this opinion.

It is so ordered.

Chief Justice Roberts, concurring. I join the opinion of the Court because I agree that under the Fifth and Sixth Amendments, a defendant is entitled to

have seen, however, Apprendi allows judges to lower sentences based on their own factual fndings; it only prohibits judges from using their own facts to increase a defendant's exposure to punishment. See n. 1, supra. As the defense bar amici are quick to point out, that helps defendants. It seems, too, Justice Jackson has directed her fre at the wrong target. The reports and law review article cited by Justice Jackson attribute sentencing disparities to this Court's remedial decision in United States v. Booker, 543 U. S. 220, 245–246 (2005), as well as various other factors: judges, “prosecutors, law enforcement offcials, probation offcers . . . , overworked defense attorneys, and other actors involved in maintaining the court system.” K. Klein & S. Klein, A Racially Biased Obstacle Course: Apprendi Transformed the Federal Sentencing Guidelines into a Series of Judicial Obstacles; Can Shame Reduce the Racial Disparities?, 99 N. C. L. Rev. 1391, 1405, 1423 (2021). And, unlike Justice Jackson, the authors of the very article she cites “celebrate Apprendi” because, in their view, it “unequivocally” gives a “bargaining chip” to defendants to seek more favorable plea offers. Id., at 1399, and n. 33; see Blakely, 542 U. S., at 311–313. 850 ERLINGER v. UNITED STATES

Thomas, J., concurring

have a jury determine beyond a reasonable doubt whether his predicate offenses were committed on different occasions for purposes of the Armed Career Criminal Act. But as Justice Kavanaugh explains, violations of that right are subject to harmless error review. See post, at 859–861 (dis- senting opinion). The Seventh Circuit should thus consider on remand the Government's contention that the error here was harmless.

Justice Thomas, concurring. I join the Court's opinion in full because it correctly ap- plies our precedents. The fact that a defendant's prior qual- ifying offenses occurred on “occasions different from one an- other” results in an increased punishment under the Armed Career Criminal Act. 18 U. S. C. § 924(e)(1). The Sixth Amendment therefore gives criminal defendants the right to have a jury fnd that fact. See Apprendi v. New Jersey, 530 U. S. 466, 501 (2000) (Thomas, J., concurring) (“[A] `crime' includes every fact that is by law a basis for imposing or increasing punishment” (emphasis added)). In Almendarez-Torres v. United States, 523 U. S. 224 (1998), the Court created a “narrow exception” to the Sixth Amendment's general rule and allowed a judge to fnd “the fact of a prior conviction,” even though that fact increases a defendant's punishment, Alleyne v. United States, 570 U. S. 99, 111, n. 1 (2013). In this case, the Court acknowledges the sharp confict between Almendarez-Torres and the Sixth Amendment. Ante, at 837–838. And, it properly declines to extend that dubious exception to the different-occasions inquiry under § 924(e)(1). Ante, at 836–842. I continue to adhere to my view that we should revisit Almendarez-Torres and correct the “error to which I suc- cumbed” by joining that decision. Apprendi, 530 U. S., at 520 (opinion of Thomas, J.); see also Sessions v. Dimaya, 584 U. S. 148, 226 (2018) (Thomas, J., dissenting); Mathis v. United States, 579 U. S. 500, 522 (2016) (Thomas, J., con- Cite as: 602 U. S. 821 (2024) 851

Kavanaugh, J., dissenting

curring); Descamps v. United States, 570 U. S. 254, 280–281 (2013) (Thomas, J., concurring in judgment); Rangel-Reyes v. United States, 547 U. S. 1200, 1202 (2006) (Thomas, J., dissenting from denial of certiorari); Shepard v. United States, 544 U. S. 13, 27–28 (2005) (Thomas, J., concurring in part and concurring in judgment). We need not overrule Almendarez-Torres to rule in Erlinger's favor, and he did not ask us to reconsider the decision—although he agrees that it should be overruled. Tr. of Oral Arg. 5. But, we have no shortage of other cases. Each Term, criminal de- fendants fle a food of petitions “specifcally presenting this Court with opportunities to reconsider Almendarez-Torres.” Rangel-Reyes, 547 U. S., at 1202 (opinion of Thomas, J.). Today's decision demonstrates further that “[i]t is time for this Court to do its part” by granting one of those many petitions and overruling Almendarez-Torres. Rangel- Reyes, 547 U. S., at 1202. Page Justice Proof Pending Kavanaugh, Publication with whom Justice Alito joins, and with whom Justice Jackson joins except as to Part III, dissenting. To prevent and punish violent crime committed with fre- arms, Congress has enacted numerous federal laws—most notably the National Firearms Act of 1934, the Gun Control Act of 1968, and the Armed Career Criminal Act of 1984. This case concerns the Armed Career Criminal Act. As relevant here, ACCA imposes a minimum sentence on a defendant who previously was convicted of at least three violent felonies committed on different occasions—and who then, after the three prior violent felony convictions, unlaw- fully possessed a frearm. In applying ACCA's minimum sentencing requirement, the Sixth Amendment allows a judge to determine whether the defendant has three or more prior convictions and whether those convictions were for vio- lent felonies. The question in this case is whether the judge may also determine whether the defendant committed those 852 ERLINGER v. UNITED STATES

prior crimes on different occasions, or instead whether a jury must do so. In my view, this Court's precedents establish that a judge may make the different-occasions determination. Because the Court today concludes that only a jury may make the different-occasions determination, I respectfully dissent.

I Paul Erlinger has been convicted of at least 9 felonies, in- cluding at least 7 burglaries. Under federal law, he there- fore may not possess a frearm. See 18 U. S. C. § 922(g)(1). As a multiple-time convicted felon, Erlinger would have re- ceived repeated notice that he could not legally possess any frearms. Yet in 2017, the police received a tip that Erlinger had recently violated that federal law and purchased a gun. Pursuant to a search warrant, offcers searched his home and found 20 guns—16 long guns and 4 handguns—and ammuni- tion to go with them. Erlinger pled guilty to possessing a frearm as a felon, in violation of § 922(g). At sentencing, the Government argued that the Armed Career Criminal Act mandated a minimum prison sentence because Erlinger had previously been convicted of at least three violent felonies committed on different occasions. The Government presented the charging documents and plea agreement for three offenses that Erlinger committed in 1991: (i) an April 4 burglary of a pizza restaurant; (ii) an April 8 burglary of a sporting-goods store; and (iii) an April 11 burglary of another restaurant. The U. S. District Court for the Southern District of Indiana determined that Er- linger had been convicted of each of those burglaries, and that those offenses qualifed as violent felonies. Erlinger did not question the judge's authority to make those determi- nations. The District Court then concluded that Erlinger committed the three burglaries on “occasions different from one another.” § 924(e)(1). Cite as: 602 U. S. 821 (2024) 853

On appeal, despite accepting the District Court's authority to determine whether he was convicted of the three prior violent felonies, Erlinger argued that the Sixth Amendment required a jury, not a judge, to determine whether he com- mitted the felonies on different occasions. That was not a promising argument. Like all 11 other Courts of Appeals that handle federal criminal cases, the U. S. Court of Appeals for the Seventh Circuit had previously ruled that a judge may determine whether a defendant's prior felonies were committed on different occasions. Adhering to that Circuit precedent, the Seventh Circuit therefore rejected Erlinger's argument. 77 F. 4th 617, 621–622 (2023) (citing United States v. Elliott, 703 F. 3d 378, 382 (2012)). The Court today, however, rejects the unanimous conclu- sion of the 12 Courts of Appeals. The Court holds that a jury must determine whether a defendant committed his prior felonies on different occasions. I respectfully disagree. II The Sixth Amendment guarantees criminal defendants the right to a “trial” by an “impartial jury.” Relying on text and history, this Court's precedents have interpreted the Sixth Amendment to draw a clear line between (i) the facts about the present crime that a jury must decide at trial and (ii) the facts about past crimes that a judge may determine at sentencing. In a series of cases that began with Apprendi v. New Jer- sey, the Court has held that a jury must fnd facts about a defendant's present offense that alter the crime's maximum or minimum possible sentence. 530 U. S. 466, 490 (2000). But Apprendi and this Court's subsequent cases have not disturbed the longstanding sentencing rule that this Court recognized two years earlier in Almendarez-Torres v. United States: Judges may resolve questions about a defendant's past crimes—questions of recidivism—that are relevant not to the defendant's guilt for the present offense but rather to 854 ERLINGER v. UNITED STATES

the length of the defendant's sentence. 523 U. S. 224, 239, 247 (1998). A This Court's opinion in Almendarez-Torres resolves the question of whether a judge may decide if the defendant com- mitted his prior violent offenses on different occasions. In that case, the Court squarely held that either a judge or a jury may apply sentence enhancements based on “recidi- vism.” Id., at 247. Almendarez-Torres involved a statute that made it a crime for a deported noncitizen to illegally reenter the United States. Id., at 229. The maximum sentence for that crime was 2 years. Ibid. But if the noncitizen had been con- victed of certain aggravated felonies before he was deported, the maximum sentence for illegal reentry increased to 20 years. Ibid. Almendarez-Torres argued that the Constitution required “Congress to treat recidivism as an element of the offense.” Id., at 239. If so, the Government would have to prove to a jury beyond a reasonable doubt that Almendarez-Torres had been convicted of an aggravated felony before his deporta- tion. Ibid. This Court rejected that argument. Id., at 247. Recidivism, the Court stated, “is a traditional, if not the most traditional, basis for a sentencing court's increasing an offender's sentence.” Id., at 243. It is not an element of the present crime of illegal reentry. Id., at 247; see also Jones v. United States, 526 U. S. 227, 249 (1999); Graham v. West Virginia, 224 U. S. 616, 629 (1912). The Court explained that the Federal and State Govern- ments have long taken different approaches to applying recidivism sentencing enhancements. Almendarez-Torres, 523 U. S., at 246. Some provide a jury trial. Ibid. Others assign recidivism enhancements to the sentencing judge. Ibid. Given the absence of any “uniform” “tradition,” the Court concluded that the choice between those methods was left to the Legislature, not governed by “a federal constitutional Cite as: 602 U. S. 821 (2024) 855

guarantee.” Id., at 246–247. To hold “that the Constitu- tion requires that recidivism be deemed an `element' of peti- tioner's offense would mark an abrupt departure from a long- standing tradition” where a judge or a jury could determine that a defendant is a recidivist. Id., at 244. Importantly, that tradition of allowing judges to apply re- cidivism sentencing enhancements developed so as to avoid signifcant prejudice to criminal defendants. If the jury had to determine recidivism, the jury would hear highly prejudi- cial evidence about a defendant's past criminal offenses at the trial for the present crime. Id., at 235. In short, in Almendarez-Torres the Court held that the Legislature can decide how it wants recidivism enhancements to be applied—by a judge or by a jury. Id., at 246–247. ACCA mandates a minimum sentence if the defendant had three prior convictions for violent felonies committed on different occasions. Erlinger does not dispute that Almendarez-Torres allows a judge to determine whether Er- linger had three prior convictions for violent felonies. The narrow question here is whether Almendarez-Torres also allows a judge to determine whether Erlinger's three prior offenses were committed on different occasions. I would conclude that Almendarez-Torres also applies to the different-occasions inquiry. Under Almendarez-Torres, legislatures may assign to judges the task of fnding facts related to a defendant's past crimes—that is, to his “re- cidivism.” Id., at 239. That principle encompasses the different-occasions inquiry. After all, the enhancement in Almendarez-Torres applied when “the defendant was previously deported subsequent to a conviction for commission of an aggravated felony.” Id., at 235 (quotation marks omitted). To apply that enhance- ment, a judge had to fnd the date on which the defendant was convicted of the aggravated felony. The judge had to fnd that the defendant in fact was the person who had com- mitted the aggravated felony. And the judge had to fnd that the defendant had been deported after that date. 856 ERLINGER v. UNITED STATES

Like the statute in Almendarez-Torres, recidivism stat- utes often require a decisionmaker to determine the who, what, when, and where of prior offenses. The Almendarez- Torres Court had no diffculty concluding that Congress could authorize judges to make the factual fndings necessary to apply the recidivism enhancement. Id., at 246. To the extent that Almendarez-Torres left any doubt on that score, this Court recognized the case's “precise holding” the follow- ing year: The “distinctive signifcance of recidivism” distin- guishes it from elements of the present crime. Jones, 526 U. S., at 248–249. Almendarez-Tor res thus means that judges can make all recidivism determinations at sentencing. Only that rule can explain this Court's cases allowing judges to fnd recidivism-related facts. See, e. g., McNeill v. United States, 563 U. S. 816, 820, 824 (2011); Shepard v. United States, 544 U. S. 13, 20–21 (2005). ACCA's different-occasions requirement similarly requires the judge to fnd the when and where of a defendant's prior criminal acts. Therefore, under Almendarez-Torres, judges may make the different-occasions determination. Given the Court's opinion in Almendarez-Torres, it comes as no sur- prise that all 12 Courts of Appeals have concluded that judges may make the different-occasions determination. See United States v. Ivery, 427 F. 3d 69, 74–75 (CA1 2005); United States v. Santiago, 268 F. 3d 151, 156 (CA2 2001); United States v. Blair, 734 F. 3d 218, 227–228 (CA3 2013); United States v. Brown, 67 F. 4th 200, 201 (CA4 2023); United States v. Davis, 487 F. 3d 282, 288 (CA5 2007); United States v. Burgin, 388 F. 3d 177, 186 (CA6 2004); United States v. Elliott, 703 F. 3d 378, 381–383 (CA7 2012); United States v. Harris, 794 F. 3d 885, 887 (CA8 2015); United States v. Walker, 953 F. 3d 577, 580–582 (CA9 2020); United States v. Harris, 447 F. 3d 1300, 1304 (CA10 2006); United States v. Weeks, 711 F. 3d 1255, 1259 (CA11 2013); United States v. Thomas, 572 F. 3d 945, 952, n. 4 (CADC 2009). Cite as: 602 U. S. 821 (2024) 857

B In concluding that Almendarez-Torres does not authorize judges to make the different-occasions determination, the Court says that subsequent decisions of this Court have “ex- pressly delimited” the reach of Almendarez-Torres. Ante, at 838, and n. 2 (citing United States v. Haymond, 588 U. S. 634, 644, n. 3 (2019) (plurality opinion); Mathis v. United States, 579 U. S. 500, 511–512 (2016); Descamps v. United States, 570 U. S. 254, 269 (2013); Alleyne v. United States, 570 U. S. 99, 111, n. 1 (2013); Cunningham v. California, 549 U. S. 270, 282 (2007); Shepard, 544 U. S., at 24 (plurality opin- ion); Blakely v. Washington, 542 U. S. 296, 301 (2004); Ap- prendi, 530 U. S., at 490). But the cases cited by the Court do not support its claim that Almendarez-Torres has been cabined, at least not in a way that would require a jury to make the different- occasions fnding. Many of those cited cases did not involve recidivism. Both Alleyne v. United States and Apprendi v. New Jersey addressed the present crime, not prior crimes. 570 U. S., at 104; 530 U. S., at 468– 469. Neither case revisited Almendarez-Torres's analysis of the distinctive history of re- cidivism sentencing enhancements for prior offenses. In- deed, both Alleyne and Apprendi expressly declined to “re- visit” Almendarez-Torres. 570 U. S., at 111, n. 1; 530 U. S., at 490. And in analyzing Almendarez-Torres, Apprendi drew a clear line between facts about prior crimes and facts about present crimes. 530 U. S., at 488. Apprendi said as much multiple times. Id., at 474, 488–489, and n. 14, 496.1 1 The other constitutional cases that the Court cites similarly did not involve recidivism. See United States v. Haymond, 588 U. S. 634, 645– 646 (2019) (plurality opinion) (“additional conduct in violation of ” the de- fendant's supervised release); Cunningham v. California, 549 U. S. 270, 275 (2007) (facts about the defendant's “violent conduct” in committing the charged crime); Blakely v. Washington, 542 U. S. 296, 298 (2004) (defend- 858 ERLINGER v. UNITED STATES

The other cases cited by the Court did not even involve a constitutional claim. In Mathis v. United States and Des- camps v. United States, the Court interpreted the text of ACCA in order to determine whether a prior conviction con- stituted a “violent felony.” 579 U. S., at 503 (quotation marks omitted); 570 U. S., at 257 (same); see also Shepard, 544 U. S., at 15–16, 19. To be sure, the Court said that its interpretation of ACCA avoided “Sixth Amendment con- cerns.” Mathis, 579 U. S., at 511; Descamps, 570 U. S., at 267. But neither case purported to articulate any Sixth Amendment holding. Importantly, constitutional avoidance “is not a method of adjudicating constitutional questions by other means. ” Clark v. Martinez, 543 U. S. 371, 381 (2005). When the Court fags potential constitutional concerns in a statutory case, that is a far cry from the Court's defnitively resolv- ing the potential constitutional issue, let alone altering previously binding constitutional precedent. A prior constitutional-avoidance holding does not absolve the Court of the duty to address the constitutional issue head-on when it is later presented to the Court. The Court's decision today mistakenly elevates constitutional-avoidance holdings to constitutional holdings. Almendarez-Torres said what it said: The Constitution does not require a jury to make recidivism determinations. And no subsequent case of this Court has cabined the holding of Almendarez-Torres. By distinguishing the different- occasions issue from the other facts about prior crimes, today's decision mistakenly crosses the clear line that Almendarez-Torres and Apprendi drew between facts about the present offense and facts about prior offenses.2

ant acted with “deliberate cruelty” in committing the charged crime (quo- tation marks omitted)). 2 Unlike the Court's interpretation of ACCA in cases like Mathis and Descamps, moreover, the Court's new constitutional rule will apply not only to federal cases, but also to state cases. Several States have recidi- Cite as: 602 U. S. 821 (2024) 859

III Even accepting the Court's interpretation of the Sixth Amendment as to the different-occasions issue, Erlinger's sentence should be affrmed. As the Government says, any error was harmless. Tr. of Oral Arg. 45. This Court has long ruled that most constitutional errors, including Sixth Amendment errors, “can be harmless.” Washington v. Recuenco, 548 U. S. 212, 218 (2006) (quotation marks omitted). The harmless-error rule serves an impor- tant purpose. It ensures that appellate courts do not set aside convictions or sentences “for small errors or defects that have little, if any, likelihood of having changed the re- sult.” Neder v. United States, 527 U. S. 1, 19 (1999) (quota- tion marks omitted). So if a constitutional error is harmless “beyond a reasonable doubt,” the defendant's conviction and sentence should be affrmed. Id., at 18. The harmless-error rule will likely ameliorate some of the short-term problems that today's decision otherwise would cause. In any case that has not become fnal, the relevant appellate court can apply harmless-error analysis.3 In most (if not all) cases, the fact that a judge rather than a jury applied ACCA's different-occasions requirement will be harmless. Whether prior felonies occurred on different occasions under ACCA is usually a straightforward question. As this Court succinctly stated in Wooden, courts “have

vism enhancements that require judges to fnd whether the defendant committed prior crimes on different occasions. See, e. g., 42 Pa. Cons. Stat. § 9714(a)(2) (2019); R. I. Gen. Laws § 12–19–21 (2002). Those state courts, like the federal courts, have long relied on Almendarez-Torres to hold that judges may make that different-occasions determination. See, e. g., Commonwealth v. Gordon, 596 Pa. 231, 251, and n. 16, 942 A. 2d 174, 186, and n. 16 (2007); State v. Ramirez, 936 A. 2d 1254, 1269 (R. I. 2007). Those courts, too, will be surprised by today's new rule. 3 For any case that is already fnal, the Teague rule will presumably bar the defendant from raising today's new rule in collateral proceedings. See Edwards v. Vannoy, 593 U. S. 255, 258 (2021); Teague v. Lane, 489 U. S. 288, 310 (1989) (plurality opinion). 860 ERLINGER v. UNITED STATES

nearly always treated offenses as occurring on separate occa- sions if a person committed them a day or more apart, or at a signifcant distance.” Wooden v. United States, 595 U. S. 360, 370 (2022) (quotation marks omitted). So in cases like this where undisputed facts establish that the defendant's prior crimes occurred a day or more apart, it will usually be evident beyond a reasonable doubt that the failure to submit the different-occasions question to the jury had no effect on the defendant's sentence.4 This case is a good example. There can be no reasonable doubt that Erlinger committed three burglaries “on occa- sions different from one another.” 18 U. S. C. § 924(e)(1). Erlinger burglarized three separate businesses, and each burglary occurred several days apart from the other two. His crimes had different victims, different dates, and differ- ent locations. And when offered the opportunity to dispute the District Court's conclusion that he committed the three burglaries on different occasions, “Erlinger supplied no ar- gument or evidence that would cast doubt” on the District Court's analysis. 77 F. 4th, at 622. In this Court, Erlinger's argument on harmless error was not much more enlightening. When asked whether he had “a viable argument” against harmless error, Erlinger re- sponded that he could “imagine somebody” who might. Tr. of Oral Arg. 25. For example, Erlinger hypothesized that someone might commit a string of burglaries several days apart to pay a single gambling debt. Ibid. Of course, Er- linger did not argue that he actually ft that description. And regardless, no “rational jury” could think that separate burglaries days apart in different places collapse into one occasion simply because the defendant committed the bur- glaries to pay a single debt. Neder, 527 U. S., at 18.

4 Going forward, I assume that the Wooden statement about offenses committed a day or more apart or at a signifcant distance will inform the content of jury instructions. Cite as: 602 U. S. 821 (2024) 861

For those reasons, I would hold that any Sixth Amend- ment error in this case was harmless. I recognize that this Court often leaves harmless-error questions to the Court of Appeals when the issue was not addressed below. See id., at 25. But that is because harmless-error questions some- times are fact-intensive and require painstaking analysis of a large record. Here, the relevant facts are simple and un- disputed: Erlinger committed three burglaries of three dif- ferent businesses on three different days, with several days separating each burglary. I would resolve the harmless- error issue in this case now rather than subjecting the par- ties to a pointless remand to the Court of Appeals and an- other round of briefng and argument, when the Court of Appeals' decision is a foregone conclusion. The Court de- clines to do so. But the harmless-error analysis will be straightforward for the Court of Appeals on remand.

IV The Court today has not overruled Almendarez-Torres; it has simply carved out the different-occasions inquiry from the general Almendarez-Torres rule. But Justice Thomas has written separately to advocate overruling Almendarez- Torres altogether. Ante, at 851 (concurring opinion). And Erlinger agreed that “the Court should someday” overrule that precedent. Tr. of Oral Arg. 5. Applying the traditional stare decisis factors, I am strongly opposed to overruling Almendarez-Torres. The principle of stare decisis is encompassed within the “judicial Power” of Article III of the Constitution. Stare decisis “promotes the evenhanded, predictable, and consist- ent development of legal principles, fosters reliance on judi- cial decisions, and contributes to the actual and perceived integrity of the judicial process.” Payne v. Tennessee, 501 U. S. 808, 827 (1991). Of course, adherence to constitutional precedent is not and should not be absolute. See Ramos v. Louisiana, 590 U. S. 83, 116–117 (2020) (Kavanaugh, J., 862 ERLINGER v. UNITED STATES

concurring in part). But the Court requires a “special jus- tifcation” or “strong grounds” before revisiting a settled holding. Id., at 120 (quotation marks omitted). That re- quirement “disciplines jurisprudential disagreement.” A. Barrett, Precedent and Jurisprudential Disagreement, 91 Texas L. Rev. 1711, 1722 (2013). In general, when considering whether to overrule a consti- tutional precedent, the Court analyzes a variety of factors that often boil down to three basic questions. First, “is the prior decision not just wrong, but grievously or egregiously wrong?” Ramos, 590 U. S., at 121 (Kavanaugh, J., con- curring in part). Second, “has the prior decision caused significant negative jurisprudential or real-world conse- quences?” Id., at 122. And third, “would overruling the prior decision unduly upset reliance interests?” Ibid. Overruling Almendarez-Torres would require running the table on all three. As I see it, however, the argument for overruling Almendarez-Torres does not satisfy any of the three requirements. A On the merits, I believe that Almendarez-Torres is correct in light of text and history. But even if one thinks that the case is wrong, it certainly is not egregiously wrong. The Sixth Amendment textually guarantees the right to a jury trial, but its text does not specify all that the right en- tails. Instead, to defne the jury right, the Court has looked to the common law, state practices in the founding era, opin- ions and treatises written soon afterward, and this Nation's historical tradition. See Ramos, 590 U. S., at 90. When those sources supply a clear answer, we can conclude that a rule is part of the jury right enshrined in the Sixth Amend- ment and therefore cannot be changed by the legislature. See id., at 90–92. No settled historical practice or understanding establishes that a jury, rather than a judge, must determine for sentenc- ing purposes whether a defendant is a recidivist. Cite as: 602 U. S. 821 (2024) 863

Statutes that enhance repeat offenders' sentences “have a long tradition in this country that dates back to colonial times.” Parke v. Raley, 506 U. S. 20, 26 (1992). Given the prevalence of those statutes, if there were a legal consensus that a jury had to determine a defendant's recidivism, we should be able to easily locate that consensus in the States' laws. After all, as the Court reminds us, every state consti- tution ratifed in the aftermath of the Revolution guaranteed a right to a trial by jury. Ante, at 829–830. Those state constitutional rights are analogous to the right guaranteed by the Sixth Amendment. But the States have traditionally used a “wide variety of methods of dealing with” recidivism enhancements. Spen- cer v. Texas, 385 U. S. 554, 566 (1967); see also Almendarez- Torres v. United States, 523 U. S. 224, 246–247 (1998). And some States have long given judges the responsibility to fnd the facts necessary to apply a recidivism sentencing enhance- ment for past crimes. States have done so to avoid the prej- udice to the defendant that would result from telling the jury about the defendant's prior crimes. One example is South Carolina. There, in the 1830s, the State had a steep recidivism enhancement for horse thieves. State v. Smith, 8 Rich. 460 (S. C. 1832). The “Court, not the jury” decided whether the enhancement applied. Id., at 461; see also State v. Allen, 8 Rich. 448, 449 (S. C. 1832); State v. Parris, 89 S. C. 140, 141, 71 S. E. 808, 809 (1911). Louisiana similarly recognized that facts about recidivism “were not essential ingredients constituting the offense charged, upon which the jury had to pass.” State v. Hud- son, 32 La. 1052, 1053 (1880). As the Supreme Court of Lou- isiana later explained, there “is no provision in the Constitu- tion that we have been able to fnd which authorizes or requires questions of fact not pertaining to the guilt or inno- cence of a defendant to be submitted to a jury.” State v. Guidry, 169 La. 215, 224, 124 So. 832, 835 (1929). Instead, the court held that the Louisiana Legislature could decide 864 ERLINGER v. UNITED STATES

whether a judge or a jury should apply the recidivism sen- tencing enhancement. So did Alabama. See Yates v. State, 245 Ala. 490, 492, 17 So. 2d 777, 779 (1944) (It is “discretionary with the trial judge whether to impose additional punishment and to make in- quiry into that question in a supplementary manner in order to apply the increased limits”). And Kansas. See State v. Woodman, 127 Kan. 166, 172, 272 P. 132, 134 (1928) (“In this state it is no concern of the jury” whether a recidivism enhancement applies); see also Chance v. State, 195 Kan. 711, 715, 408 P. 2d 677, 681 (1965) (“It has never been the rule in Kansas” that recidivism has to be treated as an element of a crime (quotation marks omit- ted)); see also 1868 Kan. Sess. Laws pp. 380–381 (recidivism enhancement). The variation in early American practice forecloses the ar- gument that the right to have a jury apply recidivism en- hancements was traditionally understood as an inherent part of the Sixth Amendment right to trial by jury.5 Another common practice in the early 1800s confrms that legislatures could choose how to handle recidivism issues. Early state constitutions required the government to include the elements of an offense in an indictment. See, e. g., Jones v. Robbins, 74 Mass. 329, 347, 350 (1857). The prosecution also had to prove elements of the offense beyond a reason- able doubt, as the requirements for an indictment and a jury trial went hand in hand. Commonwealth v. McKie, 67 Mass. 61, 62 (1854).

5 A different constitutional rule applies if the defendant's status as a felon is an element of the present offense necessary to make the conduct criminal in the frst place, as with the prohibition on possessing a fre- arm as a felon. See 18 U. S. C. § 922(g)(1); Rehaif v. United States, 588 U. S. 225, 229–230 (2019). The issue here, by contrast, is the use of past offenses to enhance the sentence for the present offense of conviction. Cite as: 602 U. S. 821 (2024) 865

But many States, including Massachusetts and Virginia, did not include the facts about past crimes in an indictment— in other words, did not treat recidivism as an element of the offense. See 1818 Mass. Acts pp. 603–604; 1819 Va. Acts ch. 171, pp. 619–620; 1824 Me. Laws p. 1009; 1868 W. Va. Acts ch. 165, pp. 733–734. Those States allowed the government to raise and prove recidivism after the defendant had been convicted of the present crime. Those States' practices re- fected and reinforced the familiar line between a conviction for the present offense and a sentencing enhancement based on past offenses. State courts upheld that practice against constitutional challenges. In 1824, the Massachusetts Supreme Judicial Court held that even if the recidivism enhancement would have been included in the original indictment “at common law,” the “legislature had” the power to change that default rule. In re Ross, 19 Mass. 165, 171; see also State v. Gra- ham, 68 W. Va. 248, 251, 69 S. E. 1010, 1011 (1910) (The West Virginia recidivism statute “is not contrary to any constitu- tional provision”). Importantly, this Court upheld that practice as well. In Graham v. West Virginia, the Court reasoned that although “the State may properly provide for the allegation of the former conviction in the indictment,” there was “no constitu- tional mandate” to do so. 224 U. S. 616, 629 (1912). In short, the historical evidence reveals a “spectrum of state procedures” for applying recidivism-based sentence en- hancements. Spencer, 385 U. S., at 566. America has a long tradition of legislative discretion over whether a judge or jury will apply recidivism sentencing enhancements.6

6 Throughout much of American history, criminal sentencing at both the state and federal levels has often been done by judges in systems where the judges have discretion to sentence within broad ranges. And in those discretionary-sentencing systems, judges routinely determine the facts of a defendant's prior convictions and take account of a defendant's criminal 866 ERLINGER v. UNITED STATES

The different state approaches did not pop up by accident. The States had good reason to experiment with different approaches to recidivism enhancements. Why? Requiring the Government to present evidence of a defendant's past crimes to a jury at trial would undermine the right to trial “by an impartial jury” that the Sixth Amendment protects. As this Court “has long recognized, the introduction of evidence of a defendant's prior crimes risks signifcant preju- dice.” Almendarez-Torres, 523 U. S., at 235. That kind of prior-crimes evidence “is said to weigh too much with the jury and to so overpersuade them as to prejudge one with a bad general record and deny him a fair opportunity to defend against a particular charge.” Michelson v. United States, 335 U. S. 469, 476 (1948). To avoid that problem, American evidence laws have long restricted the use of prior-crimes evidence. See 1 J. Wigmore, Evidence §§ 193–194, pp. 231– 235 (1904); Fed. Rule Evid. 404(b)(1). What good would that longstanding evidence principle be if the jury always had to apply recidivism enhancements? The price of having a jury apply recidivism sentencing en- hancements would be the loss of a protection against prejudi- cial evidence that itself is deeply entrenched in American law. To avoid making defendants pay that price, some legis- latures have long given judges the responsibility of applying sentence enhancements—while reserving to juries the duty to fairly decide the defendant's guilt for the present crime. See, e. g., Hudson, 32 La., at 1053 (A judge may determine “previous convictions” because that information “might prej- udice the jury”); Woodman, 127 Kan., at 172, 272 P., at 134– 135 (“the jurors' minds should not be diverted from the ques- tion of defendant's innocence or guilt by facts concerning defendant's prior convictions of other felonies,” and it “is also fairer to defendant to keep such matters entirely away from the jury”); see also D. Sidikman, Note, The Pleading and record in imposing a sentence within the broad sentencing range author- ized by statute. Cite as: 602 U. S. 821 (2024) 867

Proof of Prior Convictions in Habitual Criminal Prosecu- tions, 33 N. Y. U. L. Rev. 210, 215 (1958) (allowing judges to apply recidivism enhancements ensures that the “jury trial as to the charged offense is conducted in a nonprejudicial atmosphere”). The distinctive tradition that governs recidi- vism enhancements for past offenses has traditionally co- existed with the general right to a jury trial for present offenses. To sum up: Since the early 1800s, some legislatures have entrusted recidivism sentencing fndings to judges, and oth- ers have required juries to make those fndings. That varia- tion refects the many countervailing interests that legisla- tures must balance on this issue—including the prejudice that results from telling a jury about a defendant's past crim- inal behavior. And that unsurprising variation makes clear that Almendarez-Torres was and remains correct. The Sixth Amendment allows a judge to determine whether the defendant should receive a recidivism sentencing enhance- ment. At a minimum, the history shows that Almendarez- Torres is not egregiously wrong.

B Suppose, however, that one thinks that Almendarez- Torres is wrong, even egregiously wrong. Even then, the two other stare decisis factors—consequences and reliance inter- ests—strongly counsel in favor of adhering to Almendarez- Torres rather than overruling it. First, Almendarez-Torres has not generated the kind of negative real-world or jurisprudential consequences that would support an overruling. By allowing judges to apply recidivism enhancements, Almendarez-Torres has ensured that defendants need not choose between (i) their right to a jury trial and (ii) their interest in keeping the details of past crimes from a jury. Consider the pre judice that overruling Almendarez- Torres would cause if a jury had to decide whether defendants 868 ERLINGER v. UNITED STATES

are eligible for every state and federal recidivism sentencing enhancement. As happened in one recent federal trial, the jury could hear a prosecutor's closing argument begin: “La- dies and gentlemen of the jury, this defendant, a gun toting, drug slinging three time convicted felon . . . .” United States v. Harrell, No. 1:22–cr–20245 (SD Fla., Mar. 6, 2023), ECF Doc. 105, p. 33. The verdict in that case? Guilty. Erlinger suggests that trial courts could avoid putting de- fendants to the choice between prejudicing the jury and hav- ing to forgo a jury trial by bifurcating the trial. Bifurcating entails holding a separate mini-trial with the jury on the re- cidivism enhancement after the jury has found the defendant guilty of a crime. But most criminal cases are tried in state court. And States remain free not to bifurcate (as do federal judges, for that matter). Bifurcated trials have been and remain “rare in our jurisprudence; they have never been compelled by this Court as a matter of constitutional law, or even as a matter of federal procedure.” Spencer, 385 U. S., at 568. Instead, the trial court ordinarily has discretion to decide whether to bifurcate, and bifurcation carries “its own costs.” United States v. Durham, 655 F. Supp. 3d 598, 615 (WD Ky. 2023). In order for overruling Almendarez-Torres to help defend- ants instead of hurting many of them, this Court would have to say as a matter of constitutional law that all 50 States and the federal courts must hold bifurcated trials in recidivism cases. Of course, that would require overruling Spencer, 385 U. S., at 568–569. I doubt that the Court would go to that extreme. So if Almendarez-Torres were overruled, a defendant who is denied a bifurcated trial could be subject to all of the harm and prejudice that results from telling a jury about the defendant's past crimes.7 7 If a State does not provide for bifurcation, a defendant's only path to avoid the prejudice from the prosecutor's parading evidence of the defend- ant's past crimes before the jury may be to seek the best plea deal possible and plead guilty—hardly a benefcial result for defendants. Cite as: 602 U. S. 821 (2024) 869

Erlinger also suggests that, to avoid prejudice, defendants can stipulate to the relevant facts about their prior crimes. Brief for Petitioner 41. But that of course makes sense only in cases where the defendant is not disputing the recidivism enhancement. Stipulation is obviously not a solution for those cases where the defendant is contesting his prior con- victions and the recidivism enhancement. In any event, it is true that this Court has interpreted the Federal Rules of Evidence to require the Government to accept a defendant's stipulation that he has a prior convic- tion. Old Chief v. United States, 519 U. S. 172, 191–192 (1997). But even under the Federal Rules as interpreted in Old Chief, the jury will still hear the stipulation—for exam- ple, that the defendant admits that he committed prior felonies. And to reiterate, most criminal cases are tried in state court. This Court's interpretation of the Federal Rules of Evidence does not require state courts to follow suit when they interpret state rules of evidence. See, e. g., State v. Ball, 99–0428, p. 5 (La. 11/30/99), 756 So. 2d 275, 278 (“We conclude that Old Chief is not controlling and decline to fol- low it”); Commonwealth v. Jemison, 626 Pa. 489, 502, 98 A. 3d 1254, 1261–1262 (2014). Moreover, defendants will suffer that prejudicial harm for little beneft. Determining whether the defendant has a prior conviction for a particular offense will generally yield an obvious answer from the record (or lack thereof) of the prior conviction. Important as the judgment of the jury usually is, I struggle to imagine a jury making a different (or at least a more accurate) fnding than a judge on the ques- tion of whether a defendant was previously convicted of a crime.8 As explained above, this case is a prime example. 8 To be clear, in a recidivism proceeding (whether at trial or sentencing), a defendant ordinarily cannot challenge the underlying validity of the prior conviction—for example, by arguing that the prior conviction was infected by error. See Parke v. Raley, 506 U. S. 20, 29–30 (1992). Chal- 870 ERLINGER v. UNITED STATES

And even if one could stretch the imagination far enough to conceive of a case in which sending the prior-conviction issue to the jury makes a difference and produces a more accurate answer, it would certainly not be a suffciently frequent oc- currence to create a pressing need to overrule precedent. Nor has Almendarez-Torres caused any jurisprudential confusion. Erlinger suggests that the logic of Apprendi un- dermines Almendarez-Torres. But Apprendi itself recon- ciled the two cases. In ruling the way that it did on present crimes, Apprendi explained at length why Almendarez- Torres raised a distinct issue about past crimes. That line—between the present crime and the past crimes—is clear and has been eminently workable. For sen- tencing purposes, a judge can fnd that a defendant was con- victed of past crimes, including the who, what, when, and where of those crimes. It is as easy as that. The Apprendi rule and Almendarez-Torres rule easily co-exist and have for 24 years co-existed as a matter of Sixth Amendment law. See Apprendi v. New Jersey, 530 U. S. 466, 487–490 (2000). And today's decision does not alter the basic ease of apply- ing Almendarez-Torres. Today's decision, although I dis- agree with it, is a clear and easily administered carveout from Almendarez-Torres. Second, as to reliance, the State and Federal Governments possess substantial reliance interests in their existing sen- tencing schemes. “Statutes that punish recidivists more se- verely than frst offenders have a long tradition in this coun- try that dates back to colonial times.” Parke, 506 U. S., at 26. By 1992, all 50 States and the Federal Government had recidivism sentencing enhancements. Id., at 26–27. State judges apply many of those state recidivism sentencing en- hancements, as federal judges do in ACCA. See, e. g., N. J.

lenges to the underlying validity of a prior conviction typically must be made in the appeal or habeas corpus proceedings regarding that prior conviction. Cite as: 602 U. S. 821 (2024) 871

Jackson, J., dissenting

Stat. Ann. § 2C:44–3(a) (West 2016); 42 Pa. Cons. Stat. § 9714(a)(2) (2019); R. I. Gen. Laws § 12–19–21 (2002). Overruling Almendarez-Torres would upend that settled practice. Legislatures across the country would have to choose among various bad options. They could undermine the longstanding limits on introducing evidence at trial of past crimes. They could jettison longstanding sentencing enhancements for recidivists. They could mandate costly and ineffcient bifurcated trials in cases with a recidivism enhancement, a fairly dramatic change to day-to-day criminal trial practice in many jurisdictions. Or they could simply enact discretionary sentencing regimes and authorize sen- tences within a broad range for most crimes, leaving to judges' discretion the choice within that range. See Ap- prendi, 530 U. S., at 490, n. 16. Any of those options would be a big change—and on top of that, several would actually be unhelpful to criminal defendants. To what end? There would be little (really nothing) to gain by overruling Almendarez-Torres now, after 26 years of settled federal and state practice relying on and easily applying it. It is diffcult to conceive of a stronger case for stare decisis than a longstanding rule with substantial sys- temic benefts, ease of application, no apparent downside for criminal defendants, and signifcant reliance interests. This Court should adhere to Almendarez-Torres's settled rule.

* * * As to the Court's conclusion that a jury must determine whether the defendant's three prior violent felonies were committed on different occasions, I respectfully dissent. Justice Jackson, dissenting. In the Armed Career Criminal Act (ACCA), 18 U. S. C. § 924(e), Congress directed sentencing judges to conduct a “multi-factored” inquiry into “a range of circumstances” to determine whether a particular defendant's criminal history 872 ERLINGER v. UNITED STATES

suggests that he is the sort of “ `revolving door' felo[n]” that ACCA was designed to target. Wooden v. United States, 595 U. S. 360, 369, 375 (2022); see also § 924(e)(1). Those kinds of fndings have historically been deemed well within the capacity of a sentencing judge. See Almendarez-Torres v. United States, 523 U. S. 224, 243–244 (1998). Today, the Court concludes that Apprendi v. New Jersey, 530 U. S. 466 (2000), must be read to suggest otherwise—i. e., that under Apprendi, for sentencing purposes, facts that relate to a de- fendant's prior crimes cannot be determined by judges but instead must be found by juries. I disagree for several rea- sons, including my overarching view that Apprendi was wrongly decided. Like many jurists and other observers before me, I do not believe that Congress exceeds its consti- tutional authority when it empowers judges to make factual determinations related to punishment and directs that a par- ticular sentencing result follow from such fndings.1 I recognize, of course, that Apprendi is a binding prece- dent of this Court, and one that “has now defned the rele- vant legal regime” for nearly a quarter century. Alleyne v. United States, 570 U. S. 99, 122 (2013) (Breyer, J., concurring in part and concurring in judgment). Given that reality, un- tangling the knots Apprendi has tied is probably infeasible at this point in our Court's jurisprudential journey. But considering the faws inherent in Apprendi's approach, I can- not join today's effort to further extend Apprendi's holding, particularly when there is a well-established recidivism ex- ception to the Apprendi rule that applies to the circum- stances of the case before us now.

1 Critiques of the Apprendi Court's misguided constitutional analysis have been covered at length elsewhere. See, e. g., Apprendi v. New Jer- sey, 530 U. S. 466, 524–536 (2000) (O'Connor, J., dissenting); id., at 559–561 (Breyer, J., dissenting); Blakely v. Washington, 542 U. S. 296, 321 (2004) (O'Connor, J., dissenting); id., at 326–327 (Kennedy, J., dissenting); id., at 340–346 (Breyer, J., dissenting); S. Bibas, Judicial Fact-Finding and Sen- tence Enhancements in a World of Guilty Pleas, 110 Yale L. J. 1097 (2001). Cite as: 602 U. S. 821 (2024) 873

I agree with Justice Kavanaugh that, all things consid- ered, the Court errs in concluding today that ACCA's occa- sions inquiry must be decided by a jury. See ante, at 853– 858 (dissenting opinion). I write separately to provide an additional critical perspective on the Apprendi doctrine— one that is informed by how sentencing has actually worked on the ground, before and after Apprendi—and to note that applying the Apprendi rule to ACCA's occasions fnding cre- ates all sorts of practical problems that are easily avoided by simply allowing judges to do what they have always done. Because the Court pushes the fawed Apprendi rule past where it needs to go, and, incidentally, establishes a proce- dural requirement that is likely impossible to implement in real life, I respectfully dissent.

I In Apprendi, this Court held that “[o]ther than the fact of a prior conviction, any fact that increases the penalty for a crime beyond the prescribed statutory maximum must be submitted to a jury, and proved beyond a reasonable doubt.” 530 U. S., at 490. For the reasons others have skillfully ar- ticulated, see n. 1, supra, and also the reasons that follow, I think the Apprendi Court was wrong to interpret the Sixth Amendment's jury-trial guarantee to limit legislatures' abil- ity to defne crimes and give judges discretion to set appro- priate punishments based on fndings of fact. Apprendi and its ilk have also needlessly hampered Congress's and state legislatures' pursuit of a fairer and more rational sentenc- ing system. A Our Constitution “protects the accused against conviction except upon proof beyond a reasonable doubt of every fact necessary to constitute the crime with which he is charged.” In re Winship, 397 U. S. 358, 364 (1970). At the outset, it is important to take note of the scope of this constitutional protection as it has traditionally been recognized and under- 874 ERLINGER v. UNITED STATES

stood: It is a protection against conviction without the nec- essary facts having been established beyond a reasonable doubt. Ibid. That is not the same as a protection against being sentenced to a certain level of punishment unless the facts that are relevant to that sentencing determination have been proved to a jury consistent with the reasonable-doubt standard. The facts that must be proved before a defendant can be convicted are often called elements. See United States v. Gaudin, 515 U. S. 506, 510 (1995). Traditionally, “the ele- ments of a crime are its requisite (a) conduct (act or omission to act) and (b) mental fault (except for strict liability crimes)—plus, often, (c) specifed attendant circumstances, and, sometimes, (d) a specifed result of the conduct.” 1 W. LaFave, Substantive Criminal Law § 1.8(b), p. 103, n. 14 (3d ed. 2018); see also 1 J. Ohlin, Wharton's Criminal Law § 3:1, pp. 48–49 (16th ed. 2021). As the majority correctly recog- nizes, such elemental facts have always been in the purview of the jury. See ante, at 831. The Sixth Amendment's jury- trial guarantee refects this well-established understanding of the jury's domain. See Sullivan v. Louisiana, 508 U. S. 275, 277–278 (1993). Although sometimes the “determination of what elements constitute a crime . . . is subject to dispute,” Gaudin, 515 U. S., at 525 (Rehnquist, C. J., concurring), it is clear that “[o]nly the people's elected representatives in the legislature are authorized to `make an act a crime,' ” United States v. Davis, 588 U. S. 445, 451 (2019) (quoting United States v. Hudson, 7 Cranch 32, 34 (1812)). It follows that “ `[t]he defnition of the elements of a criminal offense is entrusted to the legislature.' ” Staples v. United States, 511 U. S. 600, 604 (1994) (quoting Liparota v. United States, 471 U. S. 419, 424 (1985); alteration in original). For that reason, this Court—at least until recent times—generally deferred to legislative judgments about which facts constitute elements of the offense. See McMillan v. Pennsylvania, 477 U. S. 79, Cite as: 602 U. S. 821 (2024) 875

85 (1986) (“[I]n determining what facts must be proved be- yond a reasonable doubt the . . . legislature's defnition of the elements . . . is usually dispositive”). Once a defendant has been found guilty of a crime—i.e., once a jury has made the requisite factual fndings establish- ing the elements of the crime—judges have traditionally been entrusted with substantial discretion to impose the ap- propriate sentence. See K. Stith & J. Cabranes, Fear of Judg- ing: Sentencing Guidelines in the Federal Courts 9 (1998). Indeed, at the dawn of our Republic, the very frst Congress enacted many criminal laws that prescribed a range of possi- ble punishments, leaving it to judges to determine the proper sentence. See An Act for the Punishment of Certain Crimes Against the United States, ch. 9, 1 Stat. 112–118; see also R. Little & T. Chen, The Lost History of Apprendi and the Blakely Petition for Rehearing, 17 Fed. Sentencing Rep. 69, 72 (2004).2 Signifcantly for present purposes, judges were thought to “inherently possess ample right to exercise reasonable, that is, judicial, discretion to enable them to wisely exert their authority” in deciding what punishment to impose. Ex parte United States, 242 U. S. 27, 41–42 (1916). In fact, a judge's determination of the appropriate sentence was long considered to be unreviewable in most circumstances. See Dorszynski v. United States, 418 U. S. 424, 431 (1974). When exercising their sentencing authority, judges were also presumed to have the power to fnd and consider nearly any fact deemed relevant to the penalty. “[B]oth before and

2 For example, the First Congress declared that misprision (i.e., conceal- ment) of a felony was punishable by “imprison[ment] not exceeding three years” and a “fn[e] not exceeding fve hundred dollars.” § 6, 1 Stat. 113. Stealing or falsifying court records was punishable by “imprison[ment] not exceeding seven years” and “whipp[ing] not exceeding thirty-nine stripes.” § 15, id., at 115–116. At least 14 other federal crimes enacted during this time gave judges discretion over similar sentencing ranges. See Little & Chen, 17 Fed. Sentencing Rep., at 72. 876 ERLINGER v. UNITED STATES

since the American colonies became a nation, courts in this country and in England practiced a policy under which a sen- tencing judge could exercise a wide discretion in the sources and types of evidence used to assist him in determining the kind and extent of punishment to be imposed within limits fxed by law.” Williams v. New York, 337 U. S. 241, 246 (1949). A sentencing judge might fnd, for example, that a defendant lacked remorse for his crime, or that the conduct underlying the crime was particularly heinous, and sentence the defendant accordingly. See id., at 247. All of those kinds of factual determinations were considered to be impor- tant factors for imposing the sentence that a person who had been found guilty of a criminal act would be required to serve. And none of them were thought to be subject to the Sixth Amendment's jury-trial right. By the late 19th century, sentencing schemes grew more complex, with the vast majority of States and the Federal Government adopting so-called indeterminate sentencing systems. A. Campbell, Law of Sentencing §§ 1:2–1:3, pp. 9– 10 (3d ed. 2004) (Campbell). In those jurisdictions, “[u]sing broad discretion, trial courts imposed minimum and maxi- mum [sentences] based on judicial estimates of how long it would take to rehabilitate criminal offenders,” which parole boards then used to determine when an offender would be released. Id., § 1:3, at 10. There, too, judges were given wide authority to determine an appropriate sentencing range, and to do so based on judicial fndings of fact. In fact, “judges were encouraged to weigh the character of the individual offender along with the nature of the offense when imposing sentence,” id., § 1:2, at 9, considerations that are immensely factbound. That judges rather that juries made these factual fndings was not thought to be constitutionally problematic. Critically, the nature of factfnding proceedings before a judge at sentencing was—and still is—fundamentally differ- ent from the factfnding that a jury engages in. Jury fact- Cite as: 602 U. S. 821 (2024) 877

fnding at trial “always ha[s] been hedged in by strict eviden- tiary procedural limitations.” Williams, 337 U. S., at 246. By contrast, such limitations have not, as a general matter, applied to judges when they fnd facts for sentencing pur- poses. See ibid. Instead, a sentencing judge has always been expected to consider a wide range of information— really, anything relevant to assessing the appropriate penalty—when determining a sentence. The difference between jury factfnding at trial and judi- cial factfnding for sentencing makes perfect sense. “Typi- cally, trial disputes center on particular issues of historical fact,” and juries accordingly “receive limited information and must choose from limited options to resolve disputed issues.” D. Berman & S. Bibas, Making Sentencing Sensible, 4 Ohio St. J. Crim. L. 37, 54 (2006). As a result, “[r]ules of evidence have been fashioned for criminal trials which narrowly con- fne the trial contest to evidence that is strictly relevant to the particular offense charged.” Williams, 337 U. S., at 246–247. “A sentencing judge, however, is not confned to the narrow issue of guilt. His task within fxed statutory or constitutional limits is to determine the type and extent of punishment after the issue of guilt has been determined.” Id., at 247. Thus, sentencing judges “receiv[e] a range of information about both the offense and the offender and can choose from various possible dispositions.” Berman & Bibas, 4 Ohio St. J. Crim. L., at 55. “Highly relevant—if not essential—to [a judge's] selection of an appropriate sentence is the possession of the fullest information possible concern- ing the defendant's life and characteristics.” Williams, 337 U. S., at 247. The upshot is that, traditionally, judges and juries have not only played different factfnding roles, they have also utilized different tools to carry out those duties. And far from being ill equipped to fnd facts for punishment purposes, judges have long been regarded as having both the power and the institutional competency to determine the factual bases for 878 ERLINGER v. UNITED STATES

the imposition of sentences. Again: This judicial authority has traditionally included the ability to make fndings of fact related to both an offender's characteristics and the criminal conduct at issue as necessary to determine an appropriate sentence—all while relying on a wide range of evidence. Historically, none of this was thought to confict with or usurp the jury's distinct role of determining guilt or innocence. B Over time, however, legislatures became concerned with “the almost wholly unchecked and sweeping powers . . . give[n] to judges in the fashioning of sentences.” M. Fran- kel, Criminal Sentences: Law Without Order 5 (1973). “[L]egislators . . . decried the perceived inequity of incarcer- ating some offenders longer than others for the same crime,” as well as the possibility that discriminatory considerations such as race and sex were playing a role in judges' sentenc- ing determinations. Campbell § 1:3, at 11; see also S. Breyer, The Federal Sentencing Guidelines and the Key Compromises Upon Which They Rest, 17 Hofstra L. Rev. 1, 4–5 (1988). “The length of time a person spent in prison appeared to depend on `what the judge ate for breakfast' on the day of sentencing, on which judge you got, or on other factors that should not have made a difference to the length of the sentence.” Blakely v. Washington, 542 U. S. 296, 332 (2004) (Breyer, J., dissenting). Out of this unregulated environment emerged a legislative development—the identifcation of what are commonly re- ferred to as “sentencing factors” (also known as “sentencing facts”). In an effort “to bring more order and consistency to the [sentencing] process,” Congress and state legislatures “sought to move from a system of indeterminate sentencing or a grant of vast discretion to the trial judge to a regime in which there [were] more uniform penalties, prescribed by the legislature.” Jones v. United States, 526 U. S. 227, Cite as: 602 U. S. 821 (2024) 879

271 (1999) (Kennedy, J., dissenting). Legislatures recog- nized that, although judges are fully competent to fnd facts and exercise discretion when sentencing, too much discre- tion could create unwarranted disparities and therefore have detrimental effects. New sentencing regimes were imple- mented to cabin sentencing discretion by “directly limit- [ing] the use . . . of particular factors in sentencing” and “by specifying statutorily how a particular factor [would] affect the sentence.” Apprendi, 530 U. S., at 560 (Breyer, J., dissenting). Legislatures sometimes specifed, for example, “that a par- ticular factor, say, use of a weapon, recidivism, injury to a victim, or bad motive, `shall' increase, or `may' increase, a particular sentence in a particular way.” Ibid. Conversely, legislatures also directed judges to disregard certain facts, including those that were deemed irrelevant for sentencing purposes. See 28 U. S. C. § 994(d) (directing the U. S. Sen- tencing Commission to consider whether age, education, vo- cational skills, and other factors are relevant to sentencing); United States Sentencing Commission, Guidelines Manual §§ 5H1.2, 5H1.4, 5H1.5, 5H1.6 (Nov. 2023) (noting that educa- tion, drug or alcohol dependence, employment record, and family ties are ordinarily not relevant in determining the length of a sentence); see also, e. g., Wash. Rev. Code § 9.94A.535(e) (2023) (excluding “[v]oluntary use of drugs or alcohol” as a potential mitigating factor). These structured sentencing schemes were not adopted “to manipulate the statutory elements of criminal offenses or to circumvent the procedural protections of the Bill of Rights.” Blakely, 542 U. S., at 316 (O'Connor, J., dissenting). Rather, Congress and the States that adopted these rules did so against a backdrop of unbounded judicial discretion that had proved, in their view, to be unwieldy, unfair, and unwise. Accordingly, the goal of legislative efforts in this regard was to constrain judicial discretion by channeling the 880 ERLINGER v. UNITED STATES

accepted competency of judges to set appropriate sentences toward the objective of achieving more consistent and more equitable outcomes. This Court dealt a signifcant blow to these legislative at- tempts to promote fairness and consistency in sentencing with its decision in Apprendi. As I previously noted, we concluded—for the frst time in history—that “[o]ther than the fact of a prior conviction, any fact that increases the pen- alty for a crime beyond the prescribed statutory maximum” is an element that “must be submitted to a jury, and proved beyond a reasonable doubt.” 530 U. S., at 490. We later extended that rule to cover any “fnding of fact” that “alters the legally prescribed punishment so as to aggravate it,” be- cause—in the Court's view—that “fact necessarily forms a constituent part of a new offense and must be submitted to the jury.” Alleyne, 570 U. S., at 114–115. By now, the Apprendi rule has been applied to a litany of punishments other than incarceration. See ante, at 833. Through these cases, the Court has “embrace[d] a universal and seemingly bright-line rule limiting the power of Con- gress and state legislatures to defne criminal offenses and the sentences that follow from convictions thereunder.” Apprendi, 530 U. S., at 525 (O'Connor, J., dissenting). By any measure, “[t]he impact of Apprendi and its progeny has been extraordinary, disrupting sentences and prompting new [corrective] legislation across the nation.” 6 W. LaFave, J. Israel, N. King, & O. Kerr, Criminal Procedure § 26.4(i), p. 1011 (4th ed. 2015). II A I was not a Member of the Court during these develop- ments. In my view, however, the Court made a serious mis- take when it confated elements and sentencing factors in this way. As others have argued, “[t]he Court's basic error in Apprendi . . . was its failure to recognize the law's tradi- Cite as: 602 U. S. 821 (2024) 881

tional distinction between elements of a crime (facts consti- tuting the crime, typically for the jury to determine) and sentencing facts (facts affecting the sentence, often concern- ing, e. g., the manner in which the offender committed the crime, and typically for the judge to determine).” Alleyne, 570 U. S., at 122 (Breyer, J., concurring in part and concur- ring in judgment). The Sixth Amendment's jury-trial right “guarantees a jury's determination of facts that constitute the elements of a crime”—no more and no less. Id., at 123. To be fair, the principal justifcation that has been given for Apprendi's confation of elements and sentencing factors is a historical one. See, e. g., ante, at 829–830; Alleyne, 570 U. S., at 108–111 (opinion of Thomas, J.). The account that has been provided in some of the Court's opinions is that, during the founding era, “ `[o]nce the facts of the offense were determined by the jury, the judge was meant simply to im- pose the prescribed sentence.' ” Ante, at 831 (quoting United States v. Haymond, 588 U. S. 634, 642 (2019) (plural- ity opinion); alteration in original). But the accuracy of this historical account is debatable. See n. 2, supra. Scholars have suggested that, far from the simplistic picture painted by the Court in Apprendi, the historical “tradition was not uniform, suggesting that the common law had no fxed rule on the subject.” S. Bibas, Judicial Fact-Finding and Sentence Enhancements in a World of Guilty Pleas, 110 Yale L. J. 1097, 1129 (2001); see also, e. g., id., at 1123–1132; Little & Chen, 17 Fed. Sentencing Rep., at 69–70; J. Mitchell, Apprendi's Domain, 2006 S. Ct. Rev. 297, 298–299. In any event, the Constitution itself does not mention sen- tencing at all—let alone the work of courts when sentenc- ing—and it certainly “does not freeze 19th-century sentenc- ing practices into permanent law.” Apprendi, 530 U. S., at 559 (Breyer, J., dissenting). Moreover, “[a]n essential aspect of the Constitution's endurance is that it empowers the politi- cal branches to address new challenges by enacting new laws and policies.” Consumer Financial Protection Bureau v. 882 ERLINGER v. UNITED STATES

Community Financial Services Assn. of America, Ltd., 601 U. S. 416, 446 (2024) (Jackson, J., concurring). In my view, the People's elected representatives should be able to pursue new and innovative approaches to sentencing and sentencing reform “without undue interference by courts,” ibid., espe- cially given that unfair and disparate sentences are a persist- ent societal problem that the legislature is indisputably au- thorized to address. Nor is there a functional, policy-based justifcation for the constitutional rule that Apprendi and its progeny enshrined. The Court has repeatedly characterized Apprendi as pre- serving “the right of jury trial” in the past, Blakely, 542 U. S., at 305, and persists with that mantra to this day, ante, at 829–832. As the reasoning goes, because the Apprendi rule recognizes that it is “ `unconstitutional for a legislature to remove from the jury the assessment of facts that increase the prescribed range of penalties to which a criminal defend- ant is exposed,' ” 530 U. S., at 490, Apprendi “preserves the historic role of the jury as an intermediary between the State and criminal defendants,” Alleyne, 570 U. S., at 114. But, in reality, the Apprendi rule does no such thing. A sentencing judge today remains free, consistent with Ap- prendi, to impose any punishment within a prescribed sen- tencing range based on whatever facts she deems relevant. See 530 U. S., at 481 (conceding that judges may “exercise discretion—taking into consideration various factors relating both to offense and offender—in imposing a judgment within the range prescribed” (emphasis deleted)). So, “[u]nder the Apprendi doctrine, the jury plays only one role with respect to sentencing, and it is an indirect one: [I]f the defendant does not plead guilty, then the jury must determine the pres- ence or absence of the verdict facts that operate to constrain the outer limit of the judge's authority to impose sentence.” B. Priester, Apprendi Land Becomes Bizarro World: “Policy Nullifcation” and Other Surreal Doctrines in the New Con- stitutional Law of Sentencing, 51 Santa Clara L. Rev. 1, 47 Cite as: 602 U. S. 821 (2024) 883

(2011) (emphasis added). Meanwhile, the judge continues to be the sole decisionmaker with respect to determining the facts she will rely upon to sentence within the typically broad statutory sentencing range. See United States v. Booker, 543 U. S. 220, 246 (2005). “The jury plays no role in extraverdict factfnding, nor in calculating the specifc sen- tence to be imposed within the outer limit authorized by the verdict facts.” Priester, 51 Santa Clara L. Rev., at 48. Apprendi's distinction between permissible and impermis- sible judicial factfnding therefore neither aligns with the doctrine's rationale nor achieves its stated goals. As a re- sult, the Apprendi rule does little actual work. Even after Apprendi, a sentencing judge can still fnd and consider any fact—including sentencing factors defned by the legisla- ture—so long as the consequence of that fact is not man- datory but rather left to the judge's discretion. And after Apprendi, just as before, criminal defendants routinely stipulate to facts that are relevant to statutory maximums and minimums as part of binding plea agreements, making factfnding with respect to these newfound elements irrele- vant. So, really, the only change that Apprendi has wrought is that legislatures may no longer limit judicial discretion as a matter of law by requiring that a particular sentencing fact have a particular effect on the sentence. See Blakely, 542 U. S., at 303–304. At bottom, then, all the Apprendi doctrine has done is “shiel[d] the sentencing power of judges from legislative en- croachment.” Priester, 51 Santa Clara L. Rev., at 49. Given this, it is no wonder that, for all its exhortations about the right to jury factfnding for sentencing purposes, the Ap- prendi line of cases appears to have had no appreciable effect on “the number of criminal jury trials” or on “the number of sentence-affecting facts decided by juries in those trials that do occur.” F. Bowman, Debacle: How the Supreme Court Has Mangled American Sentencing Law and How It Might Yet Be Mended, 77 U. Chi. L. Rev. 367, 461 (2010). 884 ERLINGER v. UNITED STATES

B In terms of the impact on the functioning of our criminal justice system, however, the consequences of the Court's de- cisions in this area have been palpable. Most notably for present purposes, Apprendi has prevented legislatures from developing innovative methods to achieve fairness in sen- tencing and thus, in my view, has stunted our collective pur- suit of justice. What I mean by this is that, while “[l]egisla- tures may set the available penalties for offenses using verdict facts,” they must now be essentially hands off “once that scope of punishment is established.” Priester, 51 Santa Clara L. Rev., at 50. Far from the mystical myth that the Sixth Amendment vests juries with sentencing power, the reality is that, through its Apprendi doctrine, the Court has merely managed to oust the legislature from its rightful place in the sentencing policy sphere, thereby effectively “in- sist[ing] that the power to consider sentencing facts and as- sess their normative worth must rest [solely] with judges.” Priester, 51 Santa Clara L. Rev., at 50. The People's representatives are left with “a binary choice” when crafting legislation due to the Apprendi doc- trine—“a fact is either of a type that triggers the full pano- ply of procedural protections that comes with the Sixth Amendment jury trial right, or it is of no constitutional con- sequence and can be found and relied on by a judge with virtually no procedural safeguards at all.” Bowman, 77 U. Chi. L. Rev., at 466; see also Blakely, 542 U. S., at 330– 340 (Breyer, J., dissenting) (outlining the limited options that legislatures have, all of which “ris[k] either impracticality, unfairness, or harm to the jury trial right”). But not every fact fts neatly into this dichotomy. Moreover, and impor- tantly, judges and juries engage with facts differently in the context of their distinct roles. “Juries provide democratic legitimacy, common sense, and fresh perspectives.” Berman & Bibas, 4 Ohio St. J. Crim. L., at 62. Meanwhile, “[j]udges are experts, can more effec- Cite as: 602 U. S. 821 (2024) 885

tively and consistently apply complex rules, and have fexi- bility in how they consider evidence.” Id., at 62–63. But under the Apprendi rule, the policymaking branches of our government can no longer devise more nuanced, creative ap- proaches to factfnding at sentencing that better refect the differing competencies of jurors and judges. In short, the Court's all-or-nothing approach to the jury- trial right in Apprendi and its kin “pose[s] a serious obstacle to [legislative] efforts to create a sentencing law that would mandate more similar treatment of like offenders, that would thereby diminish sentencing disparity, and that would conse- quently help to overcome irrational discrimination (including racial discrimination) in sentencing.” Booker, 543 U. S., at 329 (Breyer, J., dissenting in part). The Court has also “de- prive[d] Congress and state legislatures of authority that is constitutionally theirs.” Id., at 330.

C I recognize that many criminal defendants and their advo- cates prefer the Apprendi regime, which provides some de- fendants with more procedural protections at sentencing. In no way am I suggesting that the defense bar has “been `somehow duped' into advocating for a rule that would be `unfair to criminal defendants.' ” Ante, at 847 (quoting Blakely, 542 U. S., at 312). Defendants' embrace of the Ap- prendi doctrine is perfectly rational because procedural rights like the right to have a jury determine certain sen- tencing facts “hel[p] some defendants—and probably rais[e] the overall level of defense victories—by giving their law- yers claims and arguments that otherwise would not exist.” W. Stuntz, The Uneasy Relationship Between Criminal Pro- cedure and Criminal Justice, 107 Yale L. J. 1, 45 (1997). In my view, however, the beneft that some criminal de- fendants derive from the Apprendi rule in the context of their individual cases is outweighed by the negative systemic effects that Apprendi has wrought, when compared to “the 886 ERLINGER v. UNITED STATES

greater fairness of a sentencing system that a more uniform correspondence between real criminal conduct and real pun- ishment helps to create.” Blakely, 542 U. S., at 338 (Breyer, J., dissenting). An individual defendant may, of course, ben- eft from a reduced sentence based on a jury's verdict under the Apprendi rule, but that favorable outcome for one person does little to ensure systemic fairness, consistency, and trans- parency in sentencing. See Stuntz, 107 Yale L. J., at 75. The U. S. Sentencing Commission has documented the im- pact of the Apprendi rule in the wake of our decision in United States v. Booker, 543 U. S. 220, which applied Ap- prendi to the Federal Sentencing Guidelines. In Booker, we found certain judicial factfnding in the Guidelines context unconstitutional under Apprendi, but to remedy that viola- tion, we also held that the Guidelines must be treated as advisory rather than mandatory. 543 U. S., at 244–245. After Booker, sentencing disparities of all manner have in- creased signifcantly. Otherwise similarly situated defend- ants appear to receive vastly different sentences depending on the court in which they are prosecuted and what judge is assigned to their case. See, e. g., United States Sentencing Commission, Inter-District Differences in Federal Sentenc- ing Practices 7 (Jan. 2020) (“Variations in sentencing prac- tices across districts increased in the wake of the Supreme Court's 2005 decision in Booker”); United States Sentencing Commission, Intra-City Differences in Federal Sentencing Practices 7 (Jan. 2019) (“In most cities, the length of a de- fendant's sentence increasingly depends on which judge in the courthouse is assigned to his or her case”). Given the history of sentencing reform in our Nation, see Part I–B, supra, it was foreseeable that Apprendi's interference with legislative control over judicial sentencing discretion would contribute to these kinds of disparities. Among the evidence that has been amassed concerning Apprendi's negative downstream impact on sentencing fair- ness, one statistic is particularly sobering: In the federal Cite as: 602 U. S. 821 (2024) 887

criminal justice system, racial disparities in sentencing have been a persistent problem, but the gap between similarly situated Black and White male defendants “was narrowest” before the Court applied Apprendi to the Guidelines. K. Klein & S. Klein, A Racially Biased Obstacle Course: Ap- prendi Transformed the Federal Sentencing Guidelines into a Series of Judicial Obstacles; Can Shame Reduce the Racial Disparities? 99 N. C. L. Rev. 1391, 1412 (2021); see also United States Sentencing Commission, Demographic Differ- ences in Sentencing: An Update to the 2012 Booker Report 6 (Nov. 2017). And Apprendi appears to have made things appreciably worse. See Klein & Klein, 99 N. C. L. Rev., at 1412 (“Currently, for every ffty-one months a judge gives a White man, a similarly situated Black man receives eight more”). So, while the defense bar might like Apprendi because its rule can beneft individual defendants in certain cases, that rule might also be inhibiting our collective achievement of a fairer sentencing system more broadly. I do acknowledge, however, that there are risks involved with legislative innovation in this area, since not all legisla- tive action in the sentencing realm will be made in pursuit of greater systemic fairness. Legislatures are sometimes incentivized to adopt more punitive measures, such as man- datory minimums or severe recidivism-based sentencing en- hancements. See W. Stuntz, The Pathological Politics of Criminal Law, 100 Mich. L. Rev. 505, 530–531 (2001). And at least in some circumstances, the Apprendi rule operates to blunt such measures. See Alleyne, 570 U. S., at 123–124 (Breyer, J., concurring in part and concurring in judgment). But problems created by the legislature can also be ad- dressed through the democratic process; indeed, legislators have recently retreated from harsh sentencing laws. See, e. g., First Step Act of 2018, Pub. L. 115–391, 132 Stat. 5194; see also Pulsifer v. United States, 601 U. S. 124, 155 (2024) (Gorsuch, J., dissenting) (discussing the First Step Act's at- 888 ERLINGER v. UNITED STATES

tempt to “recalibrate [Congress's] approach” to sentencing). Meanwhile, Apprendi's constitutional rule operates to con- strain legislative reforms while also potentially perpetuating the unfairness caused by unwarranted disparities. To be clear, my skepticism of Apprendi should not be taken to suggest that I believe that legislatures should have unbridled authority to write laws that distinguish between sentencing factors and elements. There is, of course, “a risk of unfairness involved in permitting [legislatures] to make this labeling decision.” Blakely, 542 U. S., at 344 (Breyer, J., dissenting). Sentencing policymakers could, perhaps, “permit [a sentencing factor] to be a tail which wags the dog of the substantive offense.” McMillan, 477 U. S., at 88. For example, in the most extreme circumstances, a legisla- ture “might permit a judge to sentence an individual for mur- der though convicted only of making an illegal lane change.” Blakely, 542 U. S., at 344 (Breyer, J., dissenting). But, to me, the most logical solution to that problem is not to invoke the jury-trial right, as the Court has held. “The jury-trial right addresses only who makes certain determina- tions, not how these determinations are made.” Berman & Bibas, 4 Ohio St. J. Crim. L., at 58–59; see also Bibas, 110 Yale L. J., at 1177–1180. By contrast, “the Due Process Clause is well suited” to address unfair sentencing proce- dures. Blakely, 542 U. S., at 344 (Breyer, J., dissenting); see also Morrissey v. Brewer, 408 U. S. 471, 481 (1972) (“[D]ue process is fexible and calls for such procedural protections as the particular situation demands”). Other constitutional provisions, like the Eighth Amendment's prohibition against excessive fnes and cruel and unusual punishments, also play an obvious role in limiting the types of punishments that can be imposed based on sentencing factors. See, e. g., Ring v. Arizona, 536 U. S. 584, 619 (2002) (Breyer, J., concurring in judgment); United States v. Bajakajian, 524 U. S. 321, 334– 335 (1998). Cite as: 602 U. S. 821 (2024) 889

In other words, to my mind, not every sentencing problem is a nail requiring an Apprendi hammer. To the contrary, applying Apprendi to address these and other concerns seems to simply erect further roadblocks for policymakers who might otherwise act to promote more fairness in sentencing. III So what do my concerns about Apprendi have to do with my analysis of the question presented in this case? The doubts I have make me reluctant to join a ruling that ex- tends Apprendi's holding unnecessarily. And, here, we need not conclude that the occasions fnding in ACCA is one to which the Apprendi rule applies, as Justice Kavanaugh explains. Ante, at 853–854 (dissenting opinion). Indeed, we have already recognized that “[j]udges may,” consistent with the Apprendi rule, “resolve questions about a defend- ant's past crimes . . . that are relevant not to the defendant's guilt for the present offense but rather to the length of the defendant's sentence.” Ante, at 853–854; see also Almend- arez-Torres, 523 U. S., at 239, 247. I will use this opportunity to make one additional observa- tion: Not only is the majority's approach to ACCA's occasions fnding inconsistent with our precedent (as Justice Kava- naugh observes, ante, at 853–854), it is also unworkable in practice, due to the limitations inherent in jury presenta- tions. That is, for all the majority's talk of constitutional theory, it gives little thought to “proportionality, uniformity, and administrability,” which “are all aspects of that basic `fairness' that the Constitution demands.” Apprendi, 530 U. S., at 559 (Breyer, J., dissenting). As a reminder, ACCA directs sentencing courts to impose a 15-year mandatory minimum for a violation of 18 U. S. C. § 922(g) if the defendant has three qualifying prior con- victions “committed on occasions different from one an- other.” 18 U. S. C. § 924(e)(1). In Wooden, we explained 890 ERLINGER v. UNITED STATES

that ACCA's occasions fnding is not a simple up-or-down as- sessment; rather, the factfnder must consider “a range of circumstances,” including the timing, location, character, and relationship of the crimes. 595 U. S., at 369. The majority now concludes that, “given the intensely factual nature of this inquiry . . . , a jury must resolve it.” Ante, at 828. I come to exactly the opposite conclusion, based on the nature of the factfnding that judges and juries engage in—and their relative competencies. In my view, “the intensely factual nature of this inquiry,” ibid., when combined with the fact that ACCA's occasions fnding often pertains to long-past prior criminal behavior by the defendant, is precisely why a jury is poorly situated to make such a fnding, as opposed to a judge. As I explain below, the particular factfnding determination that the occasions inquiry requires is unsuit- able for juries to decide in terms of both fairness and eff- ciency—two crucial criteria for procedural requirements in a criminal justice system. A As to fairness, ACCA's occasions determination involves facts about a defendant's past crimes that can prejudice the jury against the defendant and thereby make it more diffcult for the jury to fnd in the defendant's favor with respect to the occasions issue. Past criminality on a defendant's part “is a traditional, if not the most traditional, basis for a sentencing court's in- creasing an offender's sentence.” Almendarez-Torres, 523 U. S., at 243. And there is a good reason judges have long been entrusted with fnding facts related to recidivism—be- cause “the introduction of evidence of a defendant's prior crimes” to a jury “risks signifcant prejudice.” Id., at 235. This Court has specifcally recognized the substantial risk of “generalizing a defendant's earlier bad act into bad character and taking that as raising the odds that he did the later bad act now charged.” Old Chief v. United States, 519 U. S. 172, 180–181 (1997). Empirical research has further confrmed Cite as: 602 U. S. 821 (2024) 891

the commonsense conclusion that criminal history is prejudi- cial. See, e. g., T. Eisenberg & V. Hans, Taking a Stand on Taking the Stand: The Effect of a Prior Criminal Record on the Decision To Testify and on Trial Outcomes, 94 Cornell L. Rev. 1353, 1357 (2009) (“Juries appear to rely on criminal records to convict when other evidence in the case normally would not support conviction”). As Justice Kavanaugh notes, such fairness concerns have long compelled courts to keep facts concerning a de- fendant's criminal history away from juries. Ante, at 866– 867. “Courts that follow the common-law tradition almost unanimously have come to disallow resort by the prosecution to any kind of evidence of a defendant's evil character to establish a probability of his guilt.” Michelson v. United States, 335 U. S. 469, 475 (1948). This principle—that juries would be prejudiced by exposure to a defendant's criminal history—is, in fact, a cornerstone of criminal procedure.3 The majority seems to agree that past-crimes evidence is prejudicial, at least to a certain extent. It says that, when ACCA's occasions inquiry is set for jury determination under the Apprendi rule (as we require today), trial courts will need “to address the prejudicial effect evidence about a de- fendant's past crimes can have on a jury.” Ante, at 847. In the majority's view, however, the “[m]ost obviou[s]” solution is bifurcating the proceedings between the § 922(g) charge and ACCA's occasions determination. Ibid. But bifurca- tion is not an easy fx, as Justice Kavanaugh explains. 3 For example, this Court has held that, when a defendant stipulates to the existence of a prior conviction, the prosecution may not introduce evi- dence of the underlying facts for that conviction. Old Chief v. United States, 519 U. S. 172, 191 (1997). Similarly, we have recognized that prosecutors may not comment on a defendant's refusal to testify, given that such a refusal may stem from a concern that his prior convictions will be used to impeach him and thus prejudice the jury. See Griffn v. California, 380 U. S. 609, 615 (1965). The suggestion that juries should now engage in rigorous factfnding with respect to a defendant's criminal history is in tension with these prior pronouncements. 892 ERLINGER v. UNITED STATES

Ante, at 868. For one thing, bifurcation of trial proceedings imposes signifcant additional burdens on the criminal justice system. Ibid. I will point to an additional problem: Even bifurcation may not suffce to completely eliminate potential juror prejudice with respect to the occasions fnding. Consider, if you will, the kinds of evidence the prosecution might ask a jury to evaluate in a bifurcated trial over (the comparatively pedestrian) question whether a defendant's prior crimes were committed on different occasions. Also imagine the defendant's potential arguments in response. Concerning the latter, in this case, Erlinger's counsel sug- gested that his past crimes—three burglaries that occurred on different dates over an 8-day period—could conceivably constitute a single occasion of criminality if those break-ins were all committed “to get money to pay [a] gambling debt.” Tr. of Oral Arg. 25. Under Wooden's inquiry, however, any jury making the occasions determination in this case would not be directed just to consider whether Erlinger did, in fact, have a gambling problem—they would also have to de- termine exactly what happened during each of Erlinger's burglaries. The jury would be called upon to assess Erlinger's credi- bility and decide whether they believed his gambling-debt story in light of his criminality, and the potential prejudice from entertaining evidence about all the sordid details of Er- linger's underlying crimes makes a fair credibility fnding much more diffcult. In other words, Erlinger's past crimi- nal behavior, and the fact that he was previously convicted of these crimes (more than one of them), conclusively estab- lishes that Erlinger is a convicted serial burglar—and per- haps a violent one at that—rendering any credibility fnding in the defendant's favor signifcantly more dubious. See Shepard v. United States, 544 U. S. 13, 38 (2005) (O'Connor, J., dissenting) (observing that the “prejudice is likely to be especially strong in ACCA cases, where the relevant prior crimes are, by defnition, `violent' ”); see also Michelson, 335 Cite as: 602 U. S. 821 (2024) 893

U. S., at 476 (observing that prior-crimes evidence can “over- persuade” jurors “to prejudge [a defendant] with a bad gen- eral record and deny him a fair opportunity to defend against” the Government's arguments). To be sure, preventing undue prejudice against defendants is an important responsibility of judges, and it is certainly possible that, with the beneft of careful limiting instructions, jurors would be able to dispassionately consider evidence about the nature and extent of a defendant's past criminality only for the narrow question whether the defendant's past crimes were, in fact, committed on separate occasions. See Spencer v. Texas, 385 U. S. 554, 562 (1967). But given what Wooden calls for, it seems as though some degree of preju- dice from the sheer fact of the defendant's having been pre- viously convicted of crimes of this nature is inevitable. See Krulewitch v. United States, 336 U. S. 440, 453 (1949) (Jack- son, J., concurring) (“The naive assumption that prejudicial effects can be overcome by instructions to the jury, all prac- ticing lawyers know to be unmitigated fction” (citation omitted)). B The potential for prejudice is not the only practical prob- lem. At the same time, a jury trial—a highly regulated, in- fexible proceeding—is, by its nature, poorly equipped to deal with the fne-grained, nuanced determinations based on sometimes-decades-old evidence that are necessary to fairly adjudicate factual questions like the one that ACCA's occa- sions inquiry raises. This mismatch, too, persists even in the proposed world of bifurcated trials. Bifurcated trials or no, it is wildly ineffcient for our system to try to ft the square peg of factfnding related to past criminality for sen- tencing purposes into the round hole of the existing proc- esses that govern jury determinations. As I have explained, factfnding at trial (before a jury) and factfnding at sentencing (before a judge) differ procedurally in fundamental ways. See Part I–A, supra. Jury factfnd- 894 ERLINGER v. UNITED STATES

ing is restricted and regimented, because a jury trial is “confne[d] . . . to evidence that is strictly relevant to the particular offense charged.” Williams, 337 U. S., at 247. Meanwhile, because “[a] sentencing judge . . . is not confned to the narrow issue of guilt,” she is not bound by “strict evidentiary procedural limitations”; rather, when determin- ing the appropriate sentence, a judge “exercise[s] a wide dis- cretion in the sources and types of evidence used to assist [her] in determining the kind and extent of punishment to be imposed.” Id., at 246–247. Again, ACCA's occasions determination is illustrative. Recall that Wooden requires a nuanced consideration of vari- ous factors, such as timing, location, and character of the past crimes, to determine whether those past crimes constituted separate “episodes of criminal activity.” 595 U. S., at 369. The “strict evidentiary procedural limitations” that apply to juries, Williams, 337 U. S., at 246, make it impractical for juries to conduct this kind of assessment. To take just one example, ACCA cases typically involve predicate crimes that may have occurred years—or even, as here, decades—ago. See ante, at 826–827. Erlinger's sentencing for the crime to which ACCA potentially applies took place in 2022. See App. to Pet. for Cert. 14a. The three burglaries that gave rise to the occasions issue occurred 31 years prior—in 1991. Id., at 21a. Given the rigidity of trials and the frailty of trial evidence, how—that is, based on what evidence—is a jury supposed to go about making the occasions fnding in this case? The majority today boldly relegates this particular fact- fnding task to a jury, without pausing to explain how this assignment will reasonably be accomplished in light of these practical limitations. “But the real world of criminal justice . . . can function only with the help of procedural compro- mises, particularly in respect to sentencing.” Apprendi, 530 U. S., at 555 (Breyer, J., dissenting). Even setting aside that a perfectly competent alternative participant in the criminal justice process (the judge) stands ready to do this (and can Cite as: 602 U. S. 821 (2024) 895

do it quite well, with the fexibility her role affords), I foresee many practical obstacles to jury factfnding concerning this particular recidivism fact, not the least of which is that evi- dence required to make the occasions determination with any reliability may not be in a form suitable for submission to a jury, or simply may no longer exist. Issues concerning the state of the evidence with respect to decades-old past crimes could be why, between Wooden and now, most sentencing judges have made ACCA's occa- sions fnding based solely on so-called Shepard documents, which “include judicial records, plea agreements, and collo- quies between a judge and the defendant” from the past criminal proceeding. Ante, at 839. In the context of a sen- tencing hearing, a judge—who, after all, has professional fa- miliarity with these kinds of records—can consider such doc- uments with minimal effort. Judges know how to interpret these sorts of court records. Additionally, during sentenc- ing proceedings, parties' arguments and evidence are not re- stricted, so based on what the parties fnd, as well as what evidence still exists, arguments can be made directly to the judge about whether the occasions inquiry is satisfed, includ- ing arguments that speak directly to imperfect recordkeep- ing and any potentially material gaps. Not so for a jury trial—at least not easily. There is a good reason why lawyers present live witnesses to juries: Showing the cold record documenting an event to a jury has much less value. At a minimum, a jury tasked with making the occasions fnding would likely need an explanation of what the Shepard documents say, and in this adversarial con- text, that explanation could probably not be provided by the lawyers on their own—it would most likely have to take the form of witness testimony. And if we were to authorize juries to go beyond the Shep- ard documents related to past crimes in order to make the occasions fnding, that inquiry would probably be even more diffcult to conduct reliably. Why? First, because, presum- ably, the original evidence and witnesses related to a defend- 896 ERLINGER v. UNITED STATES

ant's past crimes would have to be somehow located and produced, despite the passage of time and potential chain-of- custody issues. Then, once we overcome those hurdles, the rules of evidence would most likely come into play—screen- ing out potentially probative considerations. By compari- son, a sentencing judge can account for imperfections in the evidentiary records and is permitted to consider all manner of inadmissible evidence (such as out-of-court affdavits or hearsay testimony) in order to sentence. See Williams, 337 U. S., at 246.4 Also, in terms of the most effcient use of the justice sys- tem's limited resources, any witnesses could testify in narra- tive form when appearing before a sentencing judge as a factfnder, at a hearing designated for this purpose, without requiring examination by lawyers. In a jury trial, by con- trast, such evidence would likely have to satisfy “strict evi- dentiary procedural limitations,” ibid., such as direct, cross, and redirect examination. That may prove particularly and prohibitively cumbersome for the occasions inquiry, which is one small piece of the larger sentencing puzzle. All of these practical considerations lead me to believe that insisting that juries make factual determinations about a defendant's past criminal behavior—and especially the “in- tensely factual” one at issue here, ante, at 828—is not only unwise but unworkable. Again, how will juries of today ac- tually determine what happened—and why—with respect to long-forgotten crimes of yesteryear? Who will testify about those crimes (who is still around and remembers)? And

4 Some courts have held that sentencing judges cannot consider any evi- dence other than Shepard documents when undertaking ACCA's occasions inquiry. See, e. g., United States v. Elliott, 703 F. 3d 378, 382 (CA7 2012). But, in general, those conclusions appear to rest on a misunderstanding of the scope of a judge's power to fnd recidivism facts. As I have explained, judges have long been able to make factual fndings for sentencing pur- poses by considering all manner of evidence, and they are well equipped to consider any relevant evidence in making recidivism fndings. Cite as: 602 U. S. 821 (2024) 897

where is the physical evidence that was originally used to try those cases now? These kinds of challenges present one obvious reason that, even as the Apprendi Court held that the Constitution pro- hibits judges from fnding facts “that increas[e] the penalty for a crime beyond the prescribed statutory maximum,” it also included an express exemption—“[o]ther than the fact of a prior conviction.” 530 U. S., at 490. Perhaps the Court saw ft to expressly exclude the fact of a prior conviction from its original holding in Apprendi because of the reality that requiring juries to fnd recidivism facts is simply not doable. See id., at 555 (Breyer, J., dissenting) (emphasizing “the impractical nature” of the Apprendi rule). But, hey, says today's majority, why should unrealistic ex- pectations stop the Court from nonetheless requiring this to be done? Not one to be attentive to practical realities, especially when it believes it has constitutional theory on its side, the majority now plows forward, pushing the Apprendi doctrine into the realm of facts related to recidivism, which Apprendi had excluded, and which lower courts have nearly uniformly reserved for sentencing judges in the two decades since that opinion issued. The bottom line is this: Unlike juries, judges have the competency, wherewithal, and fexibility to assess facts re- lated to defendants' past crimes and to handle, in a balanced way, the various practical problems that reliance on that kind of evidence raises. All things considered, then, com- mitting the factfnding exercise related to ACCA's occasions inquiry to judges is by far more effcient, and probably more fair to participants in the justice system overall, than requir- ing juries to make that fnding. For this reason, too, this Court should have continued to allow judges to do what they have always done and what they do best—make factual fnd- ings related to a defendant's criminal history, as Apprendi seems to permit, through its acceptance of Almendarez- Torres. 898 ERLINGER v. UNITED STATES

* * * Judges take into account all kinds of facts about a criminal offense and the defendant when sentencing—they always have, and they always will. Doing so is, in fact, how a judge goes about determining what sentence to impose in a given case. Thus, the notion that it is possible for judges to fnd facts in order to “lower” but not “increase” a defendant's sentence, ante, at 835–836, n. 1 (emphasis deleted), is a theo- retical concept that bears no relationship to how sentencing actually works in a courtroom. This might well be why, in reality, judges have continued to fnd facts that relate to the penalties they impose on crimi- nal defendants (even facts that they ultimately rely on to give a higher sentence than the defendant may have other- wise received) regardless of this Court's pronouncements purporting to vindicate defendants' constitutional rights by giving juries the responsibility to make those particular fac- tual determinations. Ultimately, then, all the Apprendi rule accomplishes on the ground is impeding legislative di- rectives to courts about the exercise of judicial discretion when sentencing—a development that, in my view, does not redound to the beneft of defendants collectively, the criminal justice system, or our democratic society. In any event, before today, recidivism facts in particular have been specifcally reserved for judges to determine; Ap- prendi itself expressly exempted the fact of a prior convic- tion from the rule it was announcing. I would not extend the Apprendi rule to cover this kind of factfnding now, espe- cially since applying Apprendi to recidivism facts creates a host of practical problems that pertain to fairness and eff- ciency. Because the Court applies the Apprendi doctrine to recidivism fndings when it did not have to do so, and also reaches that conclusion without concern for the myriad prac- tical diffculties that arise from this determination, I respect- fully dissent. Reporter’s Note

The attached opinion has been revised to refect the usual publication and citation style of the United States Reports. The revised pagination makes available the offcial United States Reports citation in advance of publication. The syllabus has been prepared by the Reporter of Decisions for the convenience of the reader and constitutes no part of the opinion of the Court. A list of counsel who argued or fled briefs in this case, and who were members of the bar of this Court at the time this case was argued, has been inserted following the syllabus. Other revisions may include adjustments to formatting, captions, citation form, and any errant punctuation. The following additional edits were made:

p. 879, line 11 from bottom: “and” is replaced with “or” p. 880, line 9: “of ” is replaced with “for” p. 898, line 8 from bottom: “of ” is inserted after “kind”