Justice O’Connor
announced the judgment of the Court and delivered the opinion of the Court with respect to Parts I, II, and III, and an opinion with respect to Parts IV and V, in which The Chief Justice, Justice Scalia, and Justice Kennedy join.
In Taylor v. Louisiana, 419 U. S. 522 (1975), this Court held that the Sixth Amendment required that the jury venire be drawn from a fair cross section of the community. The Court stated, however, that “in holding that petit juries must be drawn from a source fairly representative of the community we impose no requirement that petit juries actually chosen must mirror the community and reflect the various distinctive groups in the population. Defendants are not entitled to a jury of any particular composition.” Id., at 538. The principal question presented in this case is whether the Sixth Amendment’s fair cross section requirement should now be extended to the petit jury. Because we adopt Justice Harlan’s approach to retroactivity for cases on collateral review, we leave the resolution of that question for another dav.
I
Petitioner, a black man, was convicted by an all-white Illinois jury of three counts of attempted murder, two counts of [293]*293armed robbery, and one count of aggravated battery. During jury selection for petitioner’s trial, the prosecutor used all 10 of his peremptory challenges to exclude blacks. Petitioner’s counsel used one of his 10 peremptory challenges to exclude a black woman who was married to a police officer. After the prosecutor had struck six blacks, petitioner’s counsel moved for a mistrial. The trial court denied the motion. App. 2-3. When the prosecutor struck four more blacks, petitioner’s counsel again moved for a mistrial, arguing that petitioner was “entitled to a jury of his peers.” Id., at 3. The prosecutor defended the challenges by stating that he was trying to achieve a balance of men and women on the jury. The trial court denied the motion, reasoning' that the jury “appealed] to be a fair [one].” Id., at 4.
On appeal, petitioner argued that the prosecutor’s use of peremptory challenges denied him the right to be tried by a jury that was representative of the community. The Illinois Appellate Court rejected petitioner’s fair cross section claim. People v. Teague, 108 Ill. App. 3d 891, 895-897, 439 N. E. 2d 1066, 1069-1071 (1982). The Illinois Supreme Court denied leave to appeal, and we denied certiorari. 464 U. S. 867 (1983).
Petitioner then filed a petition for a writ of habeas corpus in the United States District Court for the Northern District of Illinois. Petitioner repeated his fair cross section claim, and argued that the opinions of several Justices concurring in, or dissenting from, the denial of certiorari in McCray v. New York, 461 U. S. 961 (1983), had invited a reexamination of Swain v. Alabama, 380 U. S. 202 (1965), which prohibited States from purposefully and systematically denying blacks the opportunity to serve on juries. He also argued, for the first time, that under Swain a prosecutor could be questioned about his use of peremptory challenges once he volunteered an explanation. The District Court, though sympathetic to petitioner’s arguments, held that it was bound by Swain and Circuit precedent. App. 5-6.
[294]*294On appeal, petitioner repeated his fair cross section claim and his McCray argument. A panel of the Court of Appeals agreed with petitioner that the Sixth Amendment’s fair cross section requirement applied to the petit jury and held that petitioner had made out a prima facie case of discrimination. A majority of the judges on the Court of Appeals voted to rehear the case en banc, and the panel opinion was vacated. United States ex rel. Teague v. Lane, 779 F. 2d 1332 (CA7 1985) (en banc) (Cudahy, J., dissenting). Rehearing was postponed until after our decision in Batson v. Kentucky, 476 U. S. 79 (1986), which overruled a portion of Swain. After Batson was decided, the Court of Appeals held that petitioner could not benefit from the rule in that case because Allen v. Hardy, 478 U. S. 255 (1986) (per curiam), had held that Batson would not be applied retroactively to cases on collateral review. 820 F. 2d 832, 834, n. 4 (CA7 1987) (en banc). The Court of Appeals also held that petitioner’s Stvain claim was procedurally barred and in any event merit-less. Id., at 834, n. 6. The Court of Appeals rejected petitioner’s fair cross section claim, holding that the fair cross section requirement was limited to the jury venire. Id., at 834-843. Judge Cudahy dissented, arguing that the fair cross section requirement should be extended to the petit jury. Id., at 844.
II
Petitioner’s first contention is that he should receive the benefit of our decision in Batson even though his conviction became final before Batson was decided. Before addressing petitioner’s argument, we think it helpful to explain how Batson modified Swain. Swam held that a “State’s purposeful or deliberate denial” to blacks of an opportunity to serve as jurors solely on account of race violates the Equal Protection Clause of the Fourteenth Amendment. 380 U. S., at 203-204. In order to establish a prima facie case of discrimination under Swam, a defendant had to demonstrate that the peremptory challenge system had been “perverted.” [295]*295A defendant could raise an inference of purposeful discrimination if he showed that the prosecutor in the county where the trial was held “in case after case, whatever the circumstances, whatever the crime and whoever the defendant or the victim may be,” has been responsible for the removal of qualified' blacks who had survived challenges for cause, with the result that no blacks ever served on petit juries. Id., at 223.
In Batson, the Court overruled that portion of Swain setting forth the evidentiary showing necessary to make out a prima facie case of racial discrimination under the Equal Protection Clause. The Court held that a defendant can establish a prima facie case by showing that he is a “member of a cognizable racial group,” that the prosecutor exercised “peremptory challenges to remove from the venire members of the defendant’s race,” and that those “facts and any other relevant circumstances raise an inference that the prosecutor used that practice to exclude the veniremen from the petit jury on account of their race.” 476 U. S., at 96. Once the defendant makes out a prima facie case of discrimination, the burden shifts to the prosecutor “to come forward with a neutral explanation for challenging black jurors.” Id., at 97.
In Allen v. Hardy, the Court held that Batson constituted an “explicit and substantial break with prior precedent” because it overruled a portion of Swain. 478 U. S., at 258. Employing the retroactivity standard of Linkletter v. Walker, 381 U. S. 618, 636 (1965), the Court concluded that the rule announced in Batson should not be applied retroactively on collateral review of convictions that became final before Batson was announced. The Court defined final to mean a case “‘where the judgment of conviction was rendered, the availability of appeal exhausted, and the time for petition for certiorari had elapsed before our decision in’ Batson . . . .” 478 U. S., at 258, n. 1 (citation omitted).
Petitioner’s conviction became final 2/> years prior to Batson, thus depriving petitioner of any benefit from the rule [296]*296announced in that case. Petitioner argues, however, that Batson should be applied retroactively to all cases pending on direct review at the time certiorari was denied in McCray because the opinions filed in McCray destroyed the preceden-tial effect of Swain. Brief for Petitioner 23. The issue in McCray and its companion cases was whether the Constitution prohibited the use of peremptory challenges to exclude members of a particular group from the jury, based on the prosecutor’s assumption that they would be biased in favor of other members of that same group. Justices Marshall and Brennan dissented from the denial of certiorari, expressing the views that Swain should be reexamined and that the conduct complained of violated a defendant’s Sixth Amendment right to be tried by an impartial jury drawn from a fair cross section of the community. 461 U. S., at 964-970. Justices Stevens, Blackmun, and Powell concurred in the denial of certiorari. They agreed that the issue was an important one, but stated that it was a “sound exercise of discretion for the Court to allow the various States to serve as laboratories in which the issue receives further study before it is addressed.” Id., at 963.
We reject the basic premise of petitioner’s argument. As we have often stated, the “denial of a writ of certiorari imports no expression of opinion upon the merits of the case.” United States v. Carver, 260 U. S. 482, 490 (1923) (Holmes, J.). Accord, Hughes Tool Co. v. Trans World Airlines, Inc., 409 U. S. 363, 366, n. 1 (1973); Brown v. Allen, 344 U. S. 443, 489-497 (1963). The “variety of considerations [that] underlie denials of the writ,” Maryland v. Baltimore Radio Show, 338 U. S. 912, 917 (1950) (opinion of Frankfurter, J.), counsels against according denials of certiorari any precedential value. Concomitantly, opinions accompanying the denial of certiorari cannot have the same effect as decisions on the merits. We find that Allen v. Hardy is dis-positive, and that petitioner cannot benefit from the rule announced in Batson.
[297]*297I — H J-H
Petitioner s second contention is that he has established a violation of the Equal Protection Clause under Siva in. Recognizing that he has not shown any systematic exclusion of blacks from petit juries in case after case, petitioner contends that when the prosecutor volunteers an explanation for the use of his peremptory challenges, Swain does not preclude an examination of the stated reasons to determine the legitimacy of the prosecutor’s motive. Brief for Petitioner 35 (citing Batson, 476 U. S., at 101, n. (White, J., concurring)). See Weathersby v. Morris, 708 F. 2d 1493, 1495-1496 (CA9 1983) (supporting petitioner’s interpretation of Swain), cert. denied, 464 U. S. 1046 (1984).
Petitioner candidly admits that he did not raise the Swain claim at trial or on direct appeal. Brief for Petitioner 38-39. Because of this failure, petitioner has forfeited review of the claim in the Illinois courts. “It is well established that ‘where an appeal was taken from a conviction, the judgment of the reviewing court is res judicata as to all issues actually raised, and those that could have been presented but were not are deemed waived.’” People v. Gaines, 105 Ill. 2d 79, 87-88, 473 N. E. 2d 868, 873 (1984) (citation omitted), cert. denied, 471 U. S. 1131 (1985). The default prevents petitioner from raising the Swain claim in collateral proceedings under the Illinois Post-Conviction Hearing Act, Ill. Rev. Stat., ch. 38, ¶ 122-1 et seq. (1987), unless fundamental fairness requires that the default be overlooked. People v. Brown, 52 Ill. 2d 227, 230, 287 N. E. 2d 663, 665 (1972).
The fundamental fairness exception is a narrow one, and has been applied in limited circumstances. Compare People v. Goerger, 52 Ill. 2d 403, 406, 288 N. E. 2d 416, 418 (1972) (improper instruction on reasonable doubt “does not constitute such fundamental unfairness as to obviate the res judicata and waiver doctrines”), with People v. Ikerd, 47 Ill. 2d 211, 212, 265 N. E. 2d 120, 121 (1970) (fundamental fairness exception applies “where the right relied on has been [298]*298recognized for the first time after the direct appeal”), and People v. Hamby, 32 Ill. 2d 291, 294-295, 205 N. E. 2d 456, 458 (1965) (fundamental fairness exception applies to claims that defendant asked counsel to raise on direct appeal). It is clear that collateral relief would be unavailable to petitioner. See People v. Beamon, 31 Ill. App. 3d 145, 145-146, 333 N. E. 2d 575, 575-576 (1975) (abstract of decision) (not invoking fundamental fairness exception and holding that Swain claim not raised on direct appeal could not be raised for the first time in collateral proceedings). As a result, petitioner has exhausted his state remedies under 28 U. S. C. § 2254(b) with respect to the Swain claim. See Engle v. Isaac, 456 U. S. 107, 125-126, n. 28 (1982); United States ex rel. Williams v. Brantley, 502 F. 2d 1383, 1385-1386 (CA7 1974).
Under Wainwright v. Sykes, 433 U. S. 72, 87-91 (1977), petitioner is barred from raising the Swain claim in a federal habeas corpus proceeding unless he can show cause for the default and prejudice resulting therefrom. See Engle v. Isaac, supra, at 113-114, 117, 124-135 (applying procedural default rule to claim that had never been raised in state court). Petitioner does not attempt to show cause for his default. Instead, he argues that the claim is not barred because it was addressed by the Illinois Appellate Court. Cf. Caldwell v. Mississippi, 472 U. S. 320, 327-328 (1985). We cannot agree with petitioner’s argument. The Illinois Appellate Court rejected petitioner’s Sixth Amendment fair cross section claim ivithout mentioning the Equal Protection Clause on which Swain was based or discussing whether Swain allows a prosecutor to be questioned about his use of peremptory challenges once he volunteers an explanation. See People v. Teague, 108 Ill. App. 3d, at 895-896, 439 N. E. 2d, at 1070. Accordingly, we hold that petitioner’s Swain claim is procedurally barred, and do not address its merits.
Our application of the procedural default rule here is consistent with Hands v. Reed, ante, at 263, which holds that a “procedural default does not bar consideration of a federal [299]*299claim on either direct or habeas review unless the last state court rendering a judgment in the case ‘clearly and expressly’ states that its judgment rests on a state procedural bar” (citations and internal quotations omitted). The rule announced in Harris v. Reed assumes that a state court has had the opportunity to address a claim that is later raised in a federal habeas proceeding. It is simply inapplicable in a case such as this one, where the claim was never presented to the state courts. See ante, at 268-270 (O’Connor, J., concurring).
IV
Petitioner’s third and final contention is that the Sixth Amendment’s fair cross section requirement applies to the petit jury. As we noted at the outset, Taylor expressly stated that the fair cross section requirement does not apply to the petit jury. See 419 U. S., at 538. Petitioner nevertheless contends that the ratio decidendi of Taylor cannot be limited to the jury venire, and he urges adoption of a new rule. Because we hold that the rule urged by petitioner should not be applied retroactively to cases on collateral review, we decline to address petitioner’s contention.
A
In the past, the Court has, without discussion, often applied a new constitutional rule of criminal procedure to the defendant in the case announcing the new rule, and has confronted the question of retroactivity later when a different defendant sought the benefit of that rule. See, e. g., Brown v. Louisiana, 447 U. S. 323 (1980) (addressing retroactivity of Burch v. Louisiana, 441 U. S. 130 (1979)); Robinson v. Neil, 409 U. S. 505 (1973) (addressing retroactivity of Waller v. Florida, 397 U. S. 387 (1970)); Stovall v. Denno, 388 U. S. 293 (1967) (addressing retroactivity of United States v. Wade, 388 U. S. 218 (1967), and Gilbert v. California, 388 U. S. 263 (1967)); Tehan v. Shott, 382 U. S. 406 (1966) (addressing retroactivity of Griffin v. California, 380 U. S. 609 [300]*300(1965)). In several cases, however, the Court has addressed the retroactivity question in the very case announcing the new rule. See Morrissey v. Brewer, 408 U. S. 471, 490 (1972); Witherspoon v. Illinois, 391 U. S. 510, 523, n. 22 (1968). These two lines of cases do not have a unifying theme, and we think it is time to clarify how the question of retroactivity should be resolved for cases on collateral review.
The question of retroactivity with regard to petitioner’s fair cross section claim has been raised only in an amicus brief. See Brief for Criminal Justice Legal Foundation as Amicus Curiae 22-24. Nevertheless, that question is not foreign to the parties, who have addressed retroactivity with respect to petitioner’s Batson claim. See Brief for Petitioner 21-32; Brief for Respondent 31-38. Moreover, our sua sponte consideration of retroactivity is far from novel. In Allen v. Hardy, we addressed the retroactivity of Batson even though that question had not been presented by the petition for certiorari or addressed by the lower courts. See 478 U. S., at 261-262 (Marshall, J., dissenting). See also Mapp v. Ohio, 367 U. S. 643, 646, n. 3 (1961) (applying exclusionary rule to the States even although such a course of action was urged only by amicus curiae).
In our view, the question “whether a decision [announcing a new rule should] be given prospective or retroactive effect should be faced at the time of [that] decision.” Mish-kin, Foreword: the High Court, the Great Writ, and the Due Process of Time and Law, 79 Harv. L. Rev. 56, 64 (1965). Cf. Bowen v. United States, 422 U. S. 916, 920 (1975) (when “issues of both retroactivity and application of constitutional doctrine are raised,” the retroactivity issue should be decided first). Retroactivity is properly treated as a threshold question, for, once a new rule is applied to the defendant in the case announcing the rule, evenhanded justice requires that it be applied retroactively to all who are similarly situated. Thus, before deciding whether the fair cross section require[301]*301ment should be extended to the petit jury, we should ask whether such a rule would be applied retroactively to the case at issue. This retroactivity determination would normally entail application of the Linkletter standard, but we believe that our approach to retroactivity for cases on collateral review requires modification.
It is admittedly often difficult to determine when a case announces a new rule, and we do not attempt to define the spectrum of what may or may not constitute a new rule for retroactivity purposes. In general, however, a case announces a new rule when it breaks new ground or imposes a new obligation on the States or the Federal Government. See, e. g., Rock v. Arkansas, 483 U. S. 44, 62 (1987) (per se rule excluding all hypnotically refreshed testimony infringes impermissibly on a criminal defendant’s right to testify on his behalf); Ford v. Wainwright, 477 U. S. 399, 410 (1986) (Eighth Amendment prohibits the execution of prisoners who are insane). To put it differently, a case announces a new rule if the result was not dictated by precedent existing at the time the defendant’s conviction became final. See generally Truesdale v. Aiken, 480 U. S. 527, 528-529 (1987) (Powell, J., dissenting). Given the strong language in Taylor and our statement in Akins v. Texas, 325 U. S. 398, 403 (1945), that “[fjairness in [jury] selection has never been held to require proportional representation of races upon a jury,” application of the fair cross section requirement to the petit jury would be a new rule.1
[302]*302Not all new rules have been uniformly treated for retro-activity purposes. Nearly a quarter of a century ago, in Linkletter, the Court attempted to set some standards by which to determine the retroactivity of new rules. The question in Linkletter was whether Mapp v. Ohio, which made the exclusionary rule applicable to the States, should be applied retroactively to cases on collateral review. The Court determined that the retroactivity of Mapp should be determined by examining the purpose of the exclusionary rule, the reliance of the States on prior law, and the effect on the administration of justice of a retroactive application of the exclusionary rule. Using that standard, the Court held that Mapp would only apply to trials commencing after that case was decided. 381 U. S., at 636-640.
The Linkletter retroactivity standard has not led to consistent results. Instead, it has been used to limit application of certain new rules to cases on direct review, other new rules only to the defendants in the cases announcing such rules, and still other new rules to cases in which trials have not yet commenced. See Desist v. United States, 394 U. S. 244, 256-267 (1969) (Harlan, J., dissenting) (citing examples). [303]*303Not surprisingly, commentators have “had a veritable field day” with the Linkletter standard, with much of the discussion being “more than mildly negative.” Beytagh, Ten Years of Non-Retroactivity: A Critique and a Proposal, 61 Va. L. Rev. 1557, 1558, and n. 3 (1975) (citing sources).
Application of the Linkletter standard led to the disparate treatment of similarly situated defendants on direct review. For example, in Miranda v. Arizona, 384 U. S. 436, 467-473 (1966), the Court held that, absent other effective measures to protect the Fifth Amendment privilege against self-incrimination, a person in custody must be warned prior to interrogation that he has certain rights, including the right to remain silent. The Court applied that new rule to the defendants in Miranda and its companion cases, and held that their convictions could not stand because they had been interrogated without the proper warnings. Id., at 491-499. In Johnson v. New Jersey, 384 U. S. 719, 733-735 (1966), the Court held, under the Linkletter standard, that Miranda would only be applied to trials commencing after that decision had been announced. Because the defendant in Johnson, like the defendants in Miranda, was on direct review of his conviction, see 384 U. S., at 721, the Court’s refusal to give Miranda retroactive effect resulted in unequal treatment of those who were similarly situated. This inequity also generated vehement criticism. See, e. g., A. Bickel, The Supreme Court and the Idea of Progress 54-57 (1978) (decrying the “plain” injustice in Johnson and suggesting that the Court should have distinguished between direct and collateral review for purposes of retroactivity).
Dissatisfied with the Linkletter standard, Justice Harlan advocated a different approach to retroactivity. He argued that new rules should always be applied retroactively to cases on direct review, but that generally they should not be applied retroactively to criminal cases on collateral review. See Mackey v. United States, 401 U. S. 667, 675 (1971) (opin[304]*304ion concurring in judgments in part and dissenting in part); Desist, 394 U. S., at 256 (dissenting opinion).
In Griffith v. Kentucky, 479 U. S. 314 (1987), we rejected as unprincipled and inequitable the Linkletter standard for cases pending on direct review at the time a new rule is announced, and adopted the first part of the retroactivity approach advocated by Justice Harlan. We agreed with Justice Harlan that “failure to apply a newly declared constitutional rule to criminal cases pending on direct review violates basic norms of constitutional adjudication.” 479 U. S., at 322. We gave two reasons for our decision. First, because we can only promulgate new rules in specific cases and cannot possibly decide all cases in which review is sought, “the integrity of judicial review” requires the application of the new rule to “all similar cases pending on direct review.” Id., at 323. We quoted approvingly from Justice Harlan’s separate opinion in Mackey, supra, at 679:
“ ‘If we do not resolve all cases before us on direct review in light of our best understanding of governing constitutional principles, it is difficult to see why we should so adjudicate any case at all. ... In truth, the Court’s assertion of power to disregard current law in adjudicating cases before us that have not already run the full course of appellate review is quite simply an assertion that our constitutional function is not one of adjudication but in effect of legislation.’” 479 U. S., at 323.
Second, because “selective application of new rules violates the principle of treating similarly situated defendants the same,” we refused to continue to tolerate the inequity that resulted from not applying new rules retroactively to defendants whose cases had not yet become final. Id., at 323-324 (citing Desist, supra, at 258-259 (Harlan, J., dissenting)). Although new rules that constituted clear breaks with the past generally were not given retroactive effect under the Linkletter standard, we held that “a new rule for the conduct of criminal prosecutions is to be applied retroactively to all [305]*305cases, state or federal, pending on direct review or not yet final, with no exception for cases in which the new rule constitutes a ‘clear break’ with the past.” 479 U. S., at 328.
The Linkletter standard also led to unfortunate disparity in the treatment of similarly situated defendants on collateral review. An example will best illustrate the point. In Edwards v. Arizona, 451 U. S. 477, 484-487 (1981), the Court held that once a person invokes his right to have counsel present during custodial interrogation, a valid waiver of that right cannot be inferred from the fact that the person responded to police-initiated questioning. It was not until Solem v. Stumes, 465 U. S. 638 (1984), that the Court held, under the Linkletter standard, that Edwards was not to be applied retroactively to cases on collateral review. In the interim, several lower federal courts had come to the opposite conclusion and had applied Edwards to cases that had become final before that decision was announced. See Witt v. Wainwright, 714 F. 2d 1069, 1072-1074 (CA11 1983); Sockwell v. Maggio, 709 F. 2d 341, 343-344 (CA5 1983); McCree v. Housewright, 689 F. 2d 797, 800-802 (CA8 1982), cert. denied sub nom. McCree v. Lockhart, 460 U. S. 1088 (1983). Thus, some defendants on collateral review whose Edwards claims were adjudicated prior to Stumes received the benefit of Edwards, while those whose Edwards claims had not been addressed prior to Stumes did not. This disparity in treatment was a product of two factors: our failure to treat retroactivity as a threshold question and the Linkletter standard’s inability to account for the nature and function of collateral review. Having decided to rectify the first of those inadequacies, see supra, at 300-301, we now turn to the second.
B
Justice Harlan believed that new rules generally should not be applied retroactively to cases on collateral review. He argued that retroactivity for cases on collateral review could “be responsibly [determined] only by focusing, in the first in[306]*306stance, on the nature, function, and scope of the adjudicatory process in which such cases arise. The relevant frame of reference, in other words, is not the purpose of the new rule whose benefit the [defendant] seeks, but instead the purposes for which the writ of habeas corpus is made available.” Mackey, 401 U. S., at 682 (opinion concurring in judgments in part and dissenting in part). With regard to the nature of habeas corpus, Justice Harlan wrote:
“Habeas corpus always has been a collateral remedy, providing an avenue for upsetting judgments that have become otherwise final. It is not designed as a substitute for direct review. The interest in leaving concluded litigation in a state of repose, that is, reducing the controversy to a final judgment not subject to further judicial revision, may quite legitimately be found by those responsible for defining the scope of the writ to outweigh in some, many, or most instances the competing interest in readjudicating convictions according to all legal standards in effect when a habeas petition is filed.” Id. at 682-683.
Given the “broad scope of constitutional issues cognizable on habeas,” Justice Harlan argued that it is “sounder, in adjudicating habeas petitions, generally to apply the law prevailing at the time a conviction became final than it is to seek to dispose of [habeas] cases on the basis of intervening changes in constitutional interpretation.” Id., at 689. As he had explained in Desist, “the threat of habeas serves as a necessary additional incentive for trial and appellate courts throughout the land to conduct their proceedings in a manner consistent with established constitutional standards. In order to perform this deterrence function, . . . the habeas court need only apply the constitutional standards that prevailed at the time the original proceedings took place.” 394 U. S., at 262-263. See also Stumes, 465 U. S., at 653 (Powell, J., concurring in judgment) (“Review on habeas to determine that the conviction rests upon correct application of the [307]*307law in effect at the time of the conviction is all that is required to ‘forc[e] trial and appellate courts ... to toe the constitutional mark’ ”) (citation omitted).
Justice Harlan identified only two exceptions to his general rule of nonretroactivity for cases on collateral review. First, a new rule should be applied retroactively if it places “certain kinds of primary, private individual conduct beyond the power of the criminal law-making authority to proscribe.” Mackey, 401 U. S., at 692. Second, a new rule should be applied retroactively if it requires the observance of “those procedures that . . . are ‘implicit in the concept of ordered liberty.’” Id., at 698 (quoting Palko v. Connecticut, 302 U. S. 819, 325 (1937) (Cardozo, J.)).
Last Term, in Yates v. Aiken, 484 U. S. 211 (1988), we were asked to decide whether the rule announced in Francis v. Franklin, 471 U. S. 307 (1985), should be applied to a de: fendant on collateral review at the time that case was decided. We held that Francis did not announce a new rule because it “was merely an application of the principle that governed our decision in Sandstrom v. Montana [,442 U. S. 510 (1979)], which had been decided before [the defendant’s] trial took place.” 484 U. S., at 216-217. We therefore found it unnecessary to adopt Justice Harlan’s view of retro-activity for cases on collateral review. We stated, however, that our recent decisions had noted, as had Justice Harlan, “the important distinction between direct review and collateral review.” Id., at 215. See also Pennsylvania v. Finley, 481 U. S. 551, 555 (1987) (distinguishing between direct and collateral review for purposes of Sixth Amendment right to counsel on appeal). Indeed, we have expressly reconciled some of our retroactivity decisions with Justice Harlan’s approach. See Shea v. Louisiana, 470 U. S. 51, 58, n. 4 (1985) (giving Edwards retroactive effect on direct, but not collateral, review “is fully congruent with both aspects of the approach to retroactivity propounded by Justice Harlan”).
[308]*308We agree with Justice Harlan’s description of the function of habeas corpus. “[T]he Court never has defined the scope of the writ simply by reference to a perceived need to assure that an individual accused of crime is afforded a trial free of constitutional error.” Kuhlmann v. Wilson, 477 U. S. 436, 447 (1986) (plurality opinion). Rather, we have recognized that interests of comity and finality must also be considered in determining the proper scope of habeas review. Thus, if a defendant fails to comply with state procedural rules and is barred from litigating a particular constitutional claim in state court, the claim can be considered on federal habeas only if the defendant shows cause for the default and actual prejudice resulting therefrom. See Wainwright v. Sykes, 433 U. S., at 87-91. We have declined to make the application of the procedural default rule dependent on the magnitude of the constitutional claim at issue, see Engle v. Isaac, 466 U. S., at 129, or on the State’s interest in the enforcement of its procedural rule, see Murray v. Carrier, 477 U. S. 478, 493-496 (1986).
This Court has not “always followed an unwavering line in its conclusions as to the availability of the Great Writ. Our development of the law of federal habeas corpus has been attended, seemingly, with some backing and filling.” Fay v. Noia, 372 U. S. 391, 411-412 (1963). See also Stone v. Powell, 428 U. S. 465, 475-476 (1976). Nevertheless, it has long been established that a final civil judgment entered under a given rule of law may withstand subsequent judicial change in that rule. In Chicot County Drainage District v. Baxter State Bank, 308 U. S. 371 (1940), the Court held that a judgment based on a jurisdictional statute later found to be unconstitutional could have res judicata effect. The Court based its decision in large part on finality concerns. “The actual existence of a statute, prior to such a determination [of unconstitutionality], is an operative fact and may have consequences which cannot justly be ignored. The past cannot always be erased by a new judical declaration. . . . Questions of [309]*309. . . prior determinations deemed to have finality and acted upon accordingly . . . demand examination.” Id., at 374. Accord, Rooker v. Fidelity Trust Co., 263 U. S. 413, 415 (1923) (“Unless and until . . . reversed or modified” on appeal, an erroneous constitutional decision is “an effective and conclusive adjudication”); Thompson v. Tolmie, 2 Pet. 157, 169 (1829) (errors or mistakes of court with competent jurisdiction “cannot be corrected or examined when brought up collaterally”).
These underlying considerations of finality find significant and compelling parallels in the criminal context. Application of constitutional rules not in existence at the time a conviction became final seriously undermines the principle of finality which is essential to the operation of our criminal justice system. Without finality, the criminal law is deprived of much of its deterrent effect. The fact that life and liberty are at stake in criminal prosecutions “shows only that ‘conventional notions of finality’ should not have as much place in criminal as in civil litigation, not that they should have none.” Friendly, Is Innocence Irrelevant? Collateral Attacks on Criminal Judgments, 38 U. Chi. L. Rev. 142, 150 (1970). “[I]f a criminal judgment is ever to be final, the notion of legality must at some point include the assignment of final competence to determine legality.” Bator, Finality in Criminal Law and Federal Habeas Corpus for State Prisoners, 76 Harv. L. Rev. 441, 450-451 (1963) (emphasis omitted). See also Mackey, 401 U. S., at 691 (Harlan, J., concurring in judgments in part and dissenting in part) (“No one, not criminal defendants, not the judicial system, not society as a whole is benefited by a judgment providing a man shall tentatively go to jail today, but tomorrow and every day thereafter his continued incarceration shall be subject to fresh litigation”).
As explained by Professor Mishkin:
“From this aspect, the Linkletter problem becomes not so much one of prospectivity or retroactivity of the rule but rather of the availability of collateral attack — in [310]*310[that] case federal habeas corpus — to go behind the otherwise final judgment of conviction. . . . For the potential availability of collateral attack is what created the ‘retroactivity’ problem of Linkletter in the first place; there seems little doubt that without that possibility the Court would have given short shrift to any arguments for ‘prospective limitation’ of the Mapp rule.” Foreword, 79 Harv. L. Rev., at 77-78 (footnote omitted).
See also Bender, The Retroactive Effect of an Overruling Constitutional Decision: Mapp v. Ohio, 110 U. Pa. L. Rev. 650, 655-656 (1962).
The “costs imposed upon the State[s] by retroactive application of new rules of constitutional law on habeas corpus . . . generally far outweigh the benefits of this application.” Stumes, 465 U. S., at 654 (Powell, J., concurring in judgment). In many ways the application of new rules to cases on collateral review may be more intrusive than the enjoining of criminal prosecutions, cf. Younger v. Harris, 401 U. S. 37, 43-54 (1971), for it continually forces the States to marshal resources in order to keep in prison defendants whose trials and appeals conformed to then-existing constitutional standards. Furthermore, as we recognized in Engle v. Isaac, “[s]tate courts are understandably frustrated when they faithfully apply existing constitutional law only to have a federal court discover, during a [habeas] proceeding, new constitutional commands.” 456 U. S., at 128, n. 33. See also Brown v. Allen, 344 U. S., at 534 (Jackson, J., concurring in result) (state courts cannot “anticipate, and so comply with, this Court’s due process requirements or ascertain any standards to which this Court will adhere in prescribing them”).
We find these criticisms to be persuasive, and we now adopt Justice Harlan’s view of retroactivity for cases on collateral review. Unless they fall within an exception to the general rule, new constitutional rules of criminal procedure will not be applicable to those cases which have become final before the new rules are announced.
[311]*311V
Petitioner’s conviction became final in 1983. As a result, the rule petitioner urges would not be applicable to this case, which is on collateral review, unless it would fall within an exception.
The first exception suggested by Justice Harlan — that a new rule should be applied retroactively if it places “certain kinds of primary, private individual conduct beyond the power of the criminal law-making authority to proscribe,” Mackey, 401 U. S., at 692 (opinion concurring in judgments in part and dissenting in part) — is not relevant here. Application of the fair cross section requirement to the petit jury would not accord constitutional protection to any primary activity whatsoever.
The second exception suggested by Justice Harlan — that a new rule should be applied retroactively if it requires the observance of “those procedures that . . . are ‘implicit in the concept of ordered liberty,’” id,., at 693 (quoting Palko, 302 U. S., at 325) — we apply with a modification. The language used by Justice Harlan in Mackey leaves no doubt that he meant the second exception to be reserved for watershed rules of criminal procedure:
“Typically, it should be the case that any conviction free from federal constitutional error at the time it became final, will be found, upon reflection, to have been fundamentally fair and conducted under those procedures essential to the substance of a full hearing. However, in some situations it might be that time and growth in social capacity, as well as judicial perceptions of what we can rightly demand of the adjudicatory process, will properly alter our understanding of the bedrock procedural elements that must be found to vitiate the fairness of a particular conviction. For example, such, in my view, is the case with the right to counsel at trial now held a necessary condition precedent to any conviction [312]*312for a serious crime.” 401 U. S., at 693-694 (emphasis added).
In Desist, Justice Harlan had reasoned that one of the two principal functions of habeas corpus was “to assure that no mán has been incarcerated under a procedure which creates an impermissibly large risk that the innocent will be convicted,” and concluded “from this that all ‘new’ constitutional rules which significantly improve the pre-existing fact-finding procedures are to be retroactively applied on habeas.” 394 U. S., at 262. In Mackey, Justice Harlan gave three reasons for shifting to the less defined Palko approach. First, he observed that recent precedent, particularly Kaufman v. United, States, 394 U. S. 217 (1969) (permitting Fourth Amendment claims to be raised on collateral review), led “ineluctably... to the conclusion that it is not a principal purpose of the writ to inquire whether a criminal convict did in fact commit the deed alleged.” 401 U. S., at 694. Second, he noted that cases such as Coleman v. Alabama, 399 U. S. 1 (1970) (invalidating lineup procedures in the absence of counsel), gave him reason to doubt the marginal effectiveness of claimed improvements in factfinding. 401 U. S., at 694-695. Third, he found “inherently intractable the purported distinction between those new rules that are designed to improve the factfinding process and those designed principally to further other values.” Id., at 695.
We believe it desirable to combine the accuracy element of the Desist version of the second exception with the Mackey requirement that the procedure at issue must implicate the fundamental fairness of the trial. Were we to employ the Palko test without more, we would be doing little more than importing into a very different context the terms of the debate over incorporation. Compare Duncan v. Louisiana, 391 U. S. 145, 171-193 (1968) (Harlan, J., dissenting), with Adamson v. California, 332 U. S. 46, 68-92 (1947) (Black, J., dissenting). Reviving the Palko test now, in this area of law, would be unnecessarily anachronistic. Cf. Benton v. [313]*313Maryland, 395 U. S. 784, 794-795 (1969) (overruling Palko and incorporating the Double Jeopardy Clause). Moreover, since Mackey was decided, our cases have moved in the direction of reaffirming the relevance of the likely accuracy of convictions in determining the available scope of habeas review. See, e. g., Kuhlmann v. Wilson, 477 U. S., at 454 (plurality opinion) (a successive habeas petition may be entertained only if the defendant makes a “colorable claim of factual innocence”); Murray v. Carrier, 477 U. S., at 496 (“[W]here a constitutional violation has probably resulted in the conviction of one who is actually innocent, a federal habeas court may grant the writ even in the absence of a showing of cause for the procedural default”); Stone v. Powell, 428 U. S., at 491-492, n. 31 (removing Fourth Amendment claims from the scope of federal habeas review if the State has provided a full and fair opportunity for litigation creates no danger of denying a “safeguard against compelling an innocent man to suffer an unconstitutional loss of liberty”). Finally, we believe that Justice Harlan’s concerns about the difficulty in identifying both the existence and the value of accuracy-enhancing procedural rules can be addressed by limiting the scope of the second exception to those new procedures without which the likelihood of an accurate conviction is seriously diminished.
Because we operate from the premise that such procedures would be so central to an accurate determination of innocence or guilt, we believe it unlikely that many such components of basic due process have yet to emerge. We are also of the view that such rules are “best illustrated by recalling the classic grounds for the issuance of a writ of habeas corpus —that the proceeding was dominated by mob violence; that the prosecutor knowingly made use of perjured testimony; or that the conviction was based on a confession extorted from the defendant by brutal methods.” Rose v. [314]*314Lundy, 455 U. S. 509, 544 (1982) (Stevens, J., dissenting) (footnotes omitted).2
An examination of our decision in Taylor applying the fair cross section requirement to the jury venire leads inexorably to the conclusion that adoption of the rule petitioner urges would be a far cry from the kind of absolute prerequisite to fundamental fairness that is “implicit in the concept of ordered liberty.” The requirement that the jury venire be composed of a fair cross section of the community is based on the role of the jury in our system. Because the purpose of the jury is to guard against arbitrary abuses of power by interposing the commonsense judgment of the community between the State and the defendant, the jury venire cannot be composed only of special segments of the population. “Community participation in the administration of the criminal law ... is not only consistent with our democratic heritage but is also critical to public confidence in the fairness of the criminal justice system.” Taylor, 419 U. S., at 530. But as we stated in Daniel v. Louisiana, 420 U. S. 31, 32 (1975), which held that Taylor was not to be given retroactive effect, the fair cross section requirement “[does] not [315]*315rest on the premise that every criminal trial, or any particular trial, [is] necessarily unfair because it [is] not conducted in accordance with what we determined to be the requirements of the Sixth Amendment.” Because the absence of a fair ■cross section on the jury venire does not undermine the fundamental fairness that must underlie a conviction or seriously diminish the likelihood of obtaining an accurate conviction, we conclude that a rule requiring that petit juries be composed of a fair cross section of the community would not be a “bedrock procedural element” that would be retroactively applied under the second exception we have articulated.
Were we to recognize the new rule urged by petitioner in this case, we would have to give petitioner the benefit of that new rule even though it would not be applied retroactively to others similarly situated. In the words of Justice Brennan, such an inequitable result would be “an unavoidable consequence of the necessity that constitutional adjudications not stand as mere dictum.” Stovall v. Denno, 388 U. S., at 301. But the harm caused by the failure to treat similarly situated defendants alike cannot be exaggerated: such inequitable treatment “hardly comports with the ideal of ‘administration of justice with an even hand.’” Hankerson v. North Carolina, 432 U. S. 233, 247 (1977) (Powell, J., concurring in judgment) (quoting Desist, 394 U. S., at 255 (Douglas, J., dissenting)). See also Fuller v. Alaska, 393 U. S. 80, 82 (1968) (Douglas, J., dissenting) (if a rule is applied to the defendant in the case announcing the rule, it should be applied to all others similarly situated). Our refusal to allow such disparate treatment in the direct review context led us to adopt the first part of Justice Harlan’s retro-activity approach in Griffith. “The fact that the new rule may constitute a clear break with the past has no bearing-on the ‘actual inequity that results’ when only one of many similarly situated defendants receives the benefit of the new rule.” 479 U. S., at 327-328.
[316]*316If there were no other way to avoid rendering advisory opinions, we might well agree that the inequitable treatment described above is “an insignificant cost for adherence to sound principles of decision-making.” Stovall v. Denno, 388 U. S., at 301. But there is a more principled way of dealing with the problem. We can simply refuse to announce a new rule in a given case unless the rule would be applied retroactively to the defendant in the case and to all others similarly situated. Cf. Bowen v. United States, 422 U. S., at 920 (“This Court consistently has declined to address unsettled questions regarding the scope of decisions establishing new constitutional doctrine in cases in which it holds those decisions nonretroactive. This practice is rooted in our reluctance to decide constitutional questions unnecessarily”) (citations omitted). We think this approach is a sound one. Not only does it eliminate any problems of rendering advisory opinions, it also avoids the inequity resulting from the uneven application of new rules to similarly situated defendants. We therefore hold that, implicit in the retroactivity approach we adopt today, is the principle that habeas corpus cannot be used as a vehicle to create new constitutional rules of criminal procedure unless those rules would be applied retroactively to all defendants on collateral review through one of the two exceptions we have articulated. Because a decision extending the fair cross section requirement to the petit jury would not be applied retroactively to cases on collateral review under the approach we adopt today, we do not address petitioner’s claim.
For the reasons set forth above, the judgment of the Court of Appeals is affirmed.
It is so ordered.