with whom Justice Stevens, Justice Souter, and Justice Ginsburg join, dissenting.
In Ring v. Arizona, 536 U. S. 584 (2002), this Court held that a jury, not a judge, must make the findings necessary to *359qualify a person for punishment by death. In my view, that holding amounts to a “watershed” procedural ruling that a federal habeas court must apply when considering a constitutional challenge to a “final” death sentence — i. e., a sentence that was already final on direct review when Ring was decided.
Teague v. Lane, 489 U. S. 288 (1989) (plurality opinion), sets forth the relevant retroactivity criteria.. A new procedural rule applies retroactively in habeas proceedings if the new procedure is (1) “implicit in the concept of ordered liberty,” implicating “fundamental fairness,” and (2) “central to an accurate determination of innocence or guilt,” such that its absence “creates an impermissibly large risk that the innocent will be convicted.” Id., at 311-313 (plurality opinion) (internal quotation marks omitted). In the context of a death sentence, where the matter is not one of “innocence or. guilt,” the second criterion asks whether the new procedure is “central to an accurate determination” that death is a legally appropriate punishment. Id., at 313 (emphasis added). See Sawyer v. Smith, 497 U. S. 227, 244 (1990); O’Dell v. Netherland, 521 U. S. 151, 171, n. 3 (1997) (Stevens, J., dissenting).
The majority does not deny that Ring meets the first criterion, that its holding is “implicit in the concept of ordered liberty.” Cf. Apprendi v. New Jersey, 530 U. S. 466, 499 (2000) (Scalia, J., concurring) (absent Apprendi’s rule jury trial right “has no intelligible content”); Ring, supra, at 610 (Scalia, J., concurring) (Apprendi involves the fundamental meaning of the jury trial guarantee); Blakely v. Washington, ante, at 301-302 (tracing Apprendi’s conception of the jury trial right back to Blackstone); Duncan v. Louisiana, 391 U. S. 145, 157-158 (1968) (Sixth Amendment jury trial guarantee is a “fundamental right”). Rather, the majority focuses on whether Ring meets the second criterion: Is its rule “central to an accurate determination” that death is a legally appropriate punishment? Teague, supra, at 313.
*360As I explained in my separate concurrence in Ring, I believe the Eighth Amendment demands the use of a jury in capital sentencing because a death sentence must reflect a community-based judgment that the sentence constitutes proper retribution. See 536 U. S., at 614 (opinion concurring in judgment); see also Harris v. Alabama, 513 U. S. 504, 515-526 (1995) (Stevens, J., dissenting); Spaziano v. Florida, 468 U. S. 447, 467-490 (1984) (Stevens, J., concurring in part and dissenting in part). And a jury is significantly more likely than a judge to “express the conscience of the community on the ultimate question of life or death.” Witherspoon v. Illinois, 391 U. S. 510, 519 (1968). As Justice Stevens has pointed out:
“Juries — comprised as they are of a fair cross section of the community — are more representative institutions than is the judiciary; they reflect more accurately the composition and experiences of the community as a whole, and inevitably make decisions based on community values more reliably, than can that segment of the community that is selected for service on the bench.” Spaziano, supra, at 486-487 (footnote omitted).
On this view of the matter, the right to have jury sentencing in the capital context is both a fundamental aspect of constitutional liberty and also significantly more likely to produce an accurate assessment of whether death is the appropriate punishment.
But my view is not the Ring majority’s view. The majority held only that the jury must decide whether the special aggravating factors that make the offender eligible for death are present. 536 U. S., at 603-609. And it rested its decision that a jury, not a judge, must make that determination upon the Court’s Sixth Amendment holding in Apprendi that “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.
*361In this case, the majority says that Ring’s Apprendi-related rule cannot satisfy Teague’s accuracy-enhancing requirement, for two reasons. First, it points out that for “every argument why juries are more accurate factfinders, there is another why they are less accurate.” Ante, at 356. Hence, one cannot say “confidently” that “judicial factfinding seriously diminishes accuracy.” Ibid, (emphasis in original). Second, it relies on DeStefano v. Woods, 392 U. S. 631 (1968) (per curiam), the case in which this Court considered whether Duncan v. Louisiana, supra, which extended the Sixth Amendment jury trial guarantee to the States, should apply retroactively. The Court decided that Duncan should not have retroactive effect. “If,” the majority concludes, “a trial held entirely without a jury was not impermissibly inaccurate, it is hard to see how a trial in which a judge finds only aggravating factors could be.” Ante, at 357.
The majority, however, overlooks three additional considerations that lead me to the opposite conclusion.
First, the factfinder’s role in determining the applicability of aggravating factors in a death case is a special role that can involve, not simply the finding of brute facts, but also the making of death-related, community-based value judgments. The leading single aggravator charged in Arizona, for example, requires the factfinder to decide whether the crime was committed in an “especially heinous, cruel, or depraved manner.” Ariz. Rev. Stat. Ann. § 13-703(F)(6) (West Supp. 2003); see Office of Attorney General, State of Arizona, Capital Case Commission Final Report (2002). Three of the other four Ring-affected States use a similar aggravator. See Colo. Rev. Stat. § 18-1.3-1201(5)(j) (Lexis 2003); Idaho Code § 19-2515(9)(e) (Lexis Supp. 2003); Neb. Rev. Stat. §29-2523(1)(d) (1995). Words like “especially heinous,” “cruel,” or “depraved” — particularly when asked in the context of a death sentence proceeding — require reference to community-based standards, standards that incorporate values. (Indeed, Nebraska’s standard explicitly asks the fact-*362finder to assess the defendant’s conduct in light of “ordinary standards of morality and intelligence.” Ibid.) A jury is better equipped than a judge to identify and to apply those standards accurately. See supra, at 360.
Second, Teague’s basic purpose strongly favors retroactive application of Ring’s rule. Teague’s retroactivity principles, reflect the Court’s effort to balance competing considerations. See 489 U. S., at 309-313; Mackey v. United States, 401 U. S. 667, 675 (1971) (Harlan, J., concurring in two judgments and dissenting in one); Desist v. United States, 394 U. S. 244, 256 (1969) (Harlan, J., dissenting). On the one hand, interests related to certain of the Great Writ’s basic objectives — protecting the innocent against erroneous conviction or punishment and assuring fundamentally fair procedures — favor applying a new procedural rule retroactively. Teague, supra, at 312-313; Mackey, 401 U. S., at 693-694. So too does the legal system’s commitment to “equal justice” — i. e., to “assuring] a uniformity of ultimate treatment among prisoners.” Id., at 689.
Where death-sentence-related factfinding is at issue, these considerations have unusually strong force. This Court has made clear that in a capital case “the Eighth Amendment requires a greater degree of accuracy . . . than would be true in a noncapital case.” Gilmore v. Taylor, 508 U. S. 333, 342 (1993). Hence, the risk of error that the law can tolerate is correspondingly diminished. At the same time, the “qualitative difference of death from all other punishments” — namely, its severity and irrevocability — “requires a correspondingly greater degree of scrutiny of the capital sentencing determination” than of other criminal judgments. California v. Ramos, 463 U. S. 992, 998-999 (1983); see also Spaziano, 468 U. S., at 468 (Stevens, J., concurring in part and dissenting in part) (the Eighth Amendment mandates special safeguards to ensure that death is “a justified response to a given offense”); Ake v. Oklahoma, 470 U. S. 68, *36387 (1985) (Burger, C. J., concurring in judgment) (“In capital cases the finality of the sentence imposed warrants protections that may or may not be required in other cases”).
Consider, too, the law’s commitment to uniformity. Mackey, supra, at 689. Is treatment “uniform” when two offenders each have been sentenced to death through the-use of procedures that we now know violate the Constitution— but one is allowed to go to his death while the other receives a new, constitutionally proper sentencing proceeding? Outside the capital sentencing context, one might understand the nature of the difference that the word “finality” implies: One prisoner is already serving a final sentence, the other’s has not yet begun. But a death sentence is different in that it seems to be, and it is, an entirely future event — an event not yet undergone by either prisoner. And in respect to that event, both prisoners are, in every important respect, in the same position. I understand there is a “finality-based” difference. But given the dramatically different nature of death, that difference diminishes in importance.
Certainly the ordinary citizen will not understand the difference. That citizen will simply witness two individuals, both sentenced through the use of unconstitutional procedures, one individual going to his death, the other saved, all through an accident of timing. How can the Court square this spectacle with what it has called the “vital importance to the defendant and to the community that any decision to impose the death sentence be, and appear to be, based on reason”? Beck v. Alabama, 447 U. S. 625, 637-638 (1980) (internal quotation marks omitted).
Justice Scalia’s observation, in his concurring opinion in Ring, underscores the point. He wrote there that “the repeated spectacle of a man’s going to his death because a judge found that an aggravating factor existed” would undermine “our people’s traditional . . . veneration for the protection of the jury in criminal cases.” 536 U. S., at 612 (emphasis in original). If that is so, it is equally so whether *364the judge found that aggravating factor before or after Ring.
On the other hand, Teague recognizes that important interests argue against, and indeed generally forbid, retroactive application of new procedural rules. These interests include the “interest in insuring that there will at some point be the certainty that comes with an end to litigation”; the desirability of assuring that “attention will ultimately be focused not on whether a conviction was free from error but rather on whether the prisoner can be restored to a useful place in the community”; and the fact that society does not have endless resources to spend upon retrials, which (where witnesses have become unavailable and other evidence stale) may well produce unreliable results. Mackey, supra, at 690-691 (internal quotation marks Omitted); see also Teague, 489 U. S., at 308-310. Comity interests and respect for state autonomy point in the same direction. See id., at 308; Engle v. Isaac, 456 U. S. 107, 128, n. 33 (1982).
Certain of these interests are unusually weak where capital sentencing proceedings are at issue. Retroactivity here, for example, would not require inordinate expenditure of state resources. A decision making Ring retroactive would affect approximately 110 individuals on death row. Court Hears Arguments in Latest Death Case, N. Y. L. J., Apr. 20, 2004, p. 5. This number, however large in absolute terms, is small compared with the approximately 1.2 million individuals presently confined in state prisons. U. S. Dept. of Justice, Bureau of Justice Statistics Bulletin, Prisoner and Jail Inmates at Midyear 2003, p. 2 (May 2004). Consequently, the impact on resources is likely to be much less than if a rule affecting the ordinary criminal process were made retroactive.
Further, where the issue is “life or death,” the concern that “attention... ultimately” should be focused “on whether the prisoner can be restored to a useful place in the community” is barely relevant. Mackey, 401 U. S., at 690 (internal *365quotation marks omitted). Finally, I believe we should discount ordinary finality interests in a death case, for those interests are comparative in nature and death-related collateral proceedings, in any event, may stretch on for many years regardless. Cf. Teague, supra, at 321, n. 3 (Stevens, J., concurring in part and concurring in judgment) (“A major reason that Justice Harlan espoused limited retroactivity in collateral proceedings was the interest in making convictions final, an interest that is wholly inapplicable to the capital sentencing context”).
Third, DeStefano v. Woods, 392 U. S. 631 (1968) (per cu-riam), fails to give the majority the support for which it hopes. DeStefano did decide that Duncan’s holding — that the Sixth Amendment jury trial right applies to the States— should not have retroactive effect. But the Court decided DeStefano before Teague. And it explicitly took into account “(a) the purpose to be served by the new standards, (b) the extent of the reliance by law enforcement authorities on the old standards, and (c) the effect on the administration of justice of a retroactive application of the new standards.” 392 U. S., at 633 (internal quotation marks omitted).
The latter two factors, “reliance” and “effect on the administration of justice,” argued strongly against retroactivity. Retroactivity there, unlike here, would have thrown the prison doors open wide — at least in Louisiana and possibly in other States as well. Id., at 634. The Court believed that the first factor — “the purpose to be served by the new standards” — also favored prospective application only. But the Court described that purpose broadly, as “preventing] arbitrariness and repression”; it recognized that some judge-only trials might have been fair; and it concluded that the values served by the jury trial guarantee “would not measurably be served by requiring retrial of all persons convicted in the past” without a jury. Id., at 633-634 (emphasis added).
*366By contrast, this case involves only a small subclass of defendants deprived of jury trial rights, the relevant harm within that subclass is more widespread, the administration of justice problem is far less serious, and the reliance interest less weighty. For these reasons, I believe the DeStefano Court would have come out differently had it been considering Ring’s rule. Insofar as DeStefano has any relevance here, it highlights the importance, when making retroactivity decisions, of taking account of the considerations that underlie Teague’s categorical rules. And, as shown above, those considerations argue in favor of retroactivity in this case. See supra, at 362-365.
As I have pointed out, the majority does not deny that Ring’s rule makes some contribution to greater accuracy. It simply is unable to say “confidently” that the absence of Ring’s rule creates an “ ‘ “impermissibly large risk” ’ ” that the death penalty was improperly imposed. Ante, at 356. For the reasons stated, I believe that the risk is one that the law need not and should not tolerate. Judged in light of Teague’s basic purpose, Ring’s requirement that a jury, and not a judge, must apply the death sentence aggravators announces a watershed rule of criminal procedure that should be applied retroactively in habeas proceedings.
I respectfully dissent.