with whom Justice Souter, Justice Ginsburg, and Justice Breyer join, dissenting.
There is an acute unfairness in permitting a State to rely on a recent conviction to establish a defendant’s future dangerousness while simultaneously permitting the State to deny that there was such a conviction when the defendant attempts to argue that he is parole ineligible and therefore not a future danger. Even the most miserly reading of the opinions in Simmons v. South Carolina, 512 U. S. 154 (1994), supports the conclusion that this petitioner was denied “one of the hallmarks of due process in our adversary system,” namely, the defendant’s right “to meet the State’s case against him.” Id., at 175 (O’Connor, J., concurring in judgment).
I
In Simmons, we held that “[w]hen the State seeks to show the defendant’s future dangerousness . . . the defendant should be allowed to bring his parole ■ ineligibility to the jury’s attention — -by way of argument by defense counsel or an instruction from the court — as a means of responding to the State’s showing of future dangerousness.” Id., at 177 (O’Connor, J., concurring in judgment). The present case falls squarely within our holding.
There is no question that the Commonwealth argued Ramdass’ future dangerousness. Ante, at 161. In doing so, it focused almost entirely on Ramdass’ extensive criminal history, emphasizing that his most recent crime spree was committed after his mandatory release on parole.1 Indeed, *183the prosecution relied upon the Domino’s Pizza robbery — the very crime Virginia has precluded Ramdass from relying upon to establish his parole ineligibility.2
There is also no question that Ramdass was denied the opportunity to inform the jury of his parole ineligibility. During the sentencing deliberations, the jury asked the following question: “[I]f the Defendant is given life, is there a possibility of parole at some time before his natural death?” App. 88. Rather than giving any kind of straightforward answer, and rather than permitting counsel to explain petitioner’s parole ineligibility, the court instructed: “[Y]ou should impose such punishment as you feel is just under the evidence .... You are not to concern yourselves with what may happen afterwards.” Id., at 91.
Finally, it is undisputed that the absence of a clear instruction made a difference. The question itself demonstrates that parole ineligibility was important to the jury, and that the jury was confused about whether a' “life” sentence truly means life — or whether it means life subject to *184the possibility of parole. See Simmons, 512 U. S., at 178 (O’Connor, J., concurring in judgment) (“[T]hat the jury in this case felt compelled to ask whether parole was available shows that the jurors did not know whether or not a life-sentenced defendant will be released from prison”). More critically, three jurors said that “if the [jury] knew that [Ramdass] would have never gotten out of prison, they would have given him life rather than death.” App. 95. Two of them stated “that would have been the result among all of [the jurors] beyond question, if they had had that information.” Ibid. But “because they weren’t told or given the answer . . . they all had a perception that he would be paroled.” Ibid.3
After we remanded for reconsideration in light of Simmons, the Virginia Supreme Court held that ease did not apply because Ramdass was not “ineligible for parole when the jury was considering his sentence.” Ramdass v. Commonwealth, 248 Va. 518, 520, 450 S. E. 2d 360, 361 (1994). The applicable Virginia statute requires three strikes for a defendant to be parole ineligible. “At the time that the jury was considering Ramdass’s penalty on January 30,1993,” the court held, Ramdass “was not ineligible for parole” because he had only two strikes against him — the Pizza Hut robbery and the instant capital murder. Ibid. Ramdass’ robbery of the Domino’s Pizza did not count as his third strike, even though the jury in that case had already found him guilty. Technically, under state law, that did not count as a “conviction,” because Virginia’s definition of “conviction” is not just a guilty verdict. Rather, a “conviction” also requires a piece of paper signed by the judge entering the verdict into *185the record. Id., at 520-521, 450 S. E. 2d, at 361. The trial judge signed the entry of the judgment in the Domino’s Pizza case 19 days after the end of the sentencing phase in Ram-dass’ capital murder proceeding. Ante, at 160. Therefore, the Virginia Supreme Court held that at the time “when the jury was considering [petitioner’s] sentence” in the capital murder ease, Ramdass was “not ineligible for parole” under state law, and thus Simmons was inapplicable.
1 — I J-H
The plurality begins by stating what it thinks is the rule established in Simmons: “The parole-ineligibility instruction is required only when, assuming the jury fixes the sentence at life, the defendant is ineligible for parole under state law.” Ante, at 166. The plurality also adds a proviso: The defendant must be parole ineligible at the time of sentencing.4 Given that understanding, the plurality says “[m]aterial differences exist between this case and Simmons.” Ante, at 167. But the differences to which the plurality points do not distinguish this case from Simmons.
The first asserted distinction is that, as the Virginia Supreme Court stated, Ramdass was not parole ineligible under state law at the time of sentencing. Ramdass might *186have become parole ineligible at some later date, but at the exact moment the jury was deliberating that was not yet so. The trouble is, that is not a fact that distinguishes Ramdass’ ease from Simmons’.
In Simmons, the relevant parole statute was S. C. Code Ann. §24-21-640 (Supp. 1998). See Simmons, 512 U. S., at 176 (O’Connor, J., concurring in judgment) (citing South Carolina parole law); see also id., at 156 (plurality opinion) (same).5 Under that statute, it was the South Carolina Board of Probation, Parole, and Pardon Services that determined a defendant’s parole eligibility — and that determination would come after the sentencing phase. Then-current South Carolina case law unambiguously stated that the eligi- ■ bility determination would not be made at trial, but by the parole board.6 Moreover, the statute required the parole board to find that the defendant’s prior convictions were not committed "pursuant to one continuous course of conduct,” and it was by no means certain that the board would ultimately reach that conclusion. In fact, in Simmons the State of South Carolina steadfastly maintained that Simmons was not truly parole ineligible at the time of his sentencing *187phase because the parole board’s determination had not yet been made.7 Therefore, the fact that parole ineligibility under state law had not been determined at the time of sentencing is simply not a fact that distinguishes Simmons from Ramdass’ ease.8
*188Perhaps recognizing that problem, the plurality shifts ground. It is not, the plurality says, “only” whether parole ineligibility under state law has been determined “at the time of sentencing,” but whether there is “no possibility” of parole eligibility at that time. Ante, at 169. In other words, the plurality says that Simmons applies when there is “conclusive proof” at the time of sentencing that the defendant will (in the future) “inevitabl[y]” be found parole ineligible. Ante, at 170,173-174. In Ramdass’ case, the plurality continues, he would not inevitably be parole ineligible, because, under Virginia law, his Domino’s Pizza robbery verdict could have been set aside under Virginia Supreme Court Rule 3A:15(b) (1999). That Rule permits a trial court to set aside a guilty verdict up to 21 days after final judgment has been entered. Ante, at 174-175.9
But again, this is not a fact that distinguishes Ramdass’ ease from Simmons’. Like Virginia, South Carolina permitted (and still permits) the court to entertain post-trial motions to set aside a verdict and such a motion could have *189been filed in Simmons’ case.10 If the availability of such a post-trial procedure makes Ramdass’ parole ineligibility less than inevitable, the same must also have been true for Simmons.11 Accordingly, the mere availability of such a procedure is not a fact that distinguishes the two cases.
In the end, though, the plurality does not really rest upon inevitability at all, nor upon the alleged lack of inevitability represented by the post-trial motion procedure. Instead, the plurality relies upon the fact that at the time of Ramdass’ sentencing phase, although the jury had rendered a guilty verdict in the Domino’s Pizza robbery case, the trial judge had not yet entered judgment on the verdict. Ante, at 160, 167, 173-174, 176. That entry of judgment would come 19 days later. Ante, at 160. The distinction is important, the plurality says, because "[a] judgment, not a verdict, is the usual measure for finality in the trial court,” ante, at 176, whereas a verdict without a judgment is “uncertain,” ibid. The plurality is, of course, correct that the missing entry of judgment is a circumstance that was not present in Simmons. *190But the plurality’s entirely unsupported assertion that an entry of judgment is more “certain” than a verdict is just flat wrong.
The sole basis for the plurality’s conclusion that the Domino’s Pizza verdict is uncertain is the possibility that it could be set aside under Rule 3A:15(b). But under that Rule, a guilty verdict may be set aside even after judgment has been entered. See n. 9, supra. The plurality has cited not a single case suggesting that the standard for setting aside a verdict under Rule 3A:15(b) varies depending on whether or not judgment has been entered. Accordingly, a verdict that is susceptible to being set aside under Rule 3A:15(b) is no more or less certain simply because judgment has been entered on that verdict; whatever the degree of uncertainty is, it is identical in both cases. In short, whether judgment has been entered on the verdict has absolutely no bearing on the verdict’s “uncertainty.”
The plurality cites 11 Virginia cases to support its argument that Rule 3A:15(b) puts a verdict on shaky ground. Ante, at 174-175. The authorities are less than overwhelming. Only 2 of those 11 cases actually mention Rule 3A:15(b),12 and one of those does so in dicta in a footnote in the unpublished decision of an intermediate state court.13 Four others make passing reference to some sort of post-trial motion that was denied, but do so only in the context of reciting the procedural history of the case under review.14 *191Another ease also makes passing reference to the denial of a post-trial motion, but it is clear from the fact that the motion was predicated on "new evidence” (which is not a basis for a Rule 3A:15(b) motion, see n. 9, supra) and was made four months after the verdict that the motion was almost certainly not based on Rule 3A:15(b).15 Ultimately, the plurality points to only three cases to demonstrate that "a jury verdict [is] uncertain and unreliable until judgment is entered.” Ante, at 175 (citing Dowell v. Commonwealth, 12 Va. App. 1145, 1149, 408 S. E. 2d 263, 265 (1991) (mentioning Rule 3A:15(b)); Smith v. Commonwealth, 134 Va. 589, 113 S. E. 707 (1922); Blair v. Commonwealth, 66 Va. 850 (1874)). What these cases hold, however, is (1) that a verdict without an entry of judgment may not be used for purposes of impeaching a witness’ credibility; (2) the same may not be used for purposes of a statute permitting the removal from public office of any person “convicted of an act... involving moral turpitude”; but (3) the Governor can pardon a prisoner after a verdict and before entry of judgment. Not one of them actually involves a Rule 3A:15(b) motion, nor remotely says that a verdict itself is “unreliable.”16 The *192plurality scrounges to find case law support, but the result barely registers on the radar screen.
Furthermore, the plurality thinks that there is “no authority” for the proposition that entry of judgment is generally considered to be a “ministerial” matter. Ante, at 174. In a related context, however, the 'Virginia Supreme Court has observed:
“The rendition of a judgment is to be distinguished from its entry in the records. The rendition of a judgment is the judicial act of the court, whereas the entry of a judgment by the clerk on the records of the court is a ministerial, and not a judicial, act.. . . The entry or recor-dation of such an instrument in an order book‘is the ministerial act of the clerk and does not constitute an integral part of the judgment.” Rollins v. Bazile, 205 Va. 613, 617, 139 S. E. 2d 114, 117 (1964) (citations and internal quotation marks omitted).
In any event, there is a more critical point to be made about the plurality’s entry-of-judgment distinction. In relying on that distinction, the plurality is necessarily abandoning the very understanding of Simmons that it purports to be following. As explained above, to the extent that the availability of Rule 3A:15(b) motions undermines the inevitability of a defendant’s prior verdicts (and therefore his parole ineligibility) under state law, it does so whether or not judgment has been entered on the verdict. So why is it that Simmons does not apply when there is no entry of judgment?
The answer simply cannot be that, under state law, and at the time of sentencing, the defendant will not inevitably be *193found parole ineligible: the inevitability of the verdict is undermined, equally with or without the judgment; and the defendant is eligible for parole under state law if the verdict is set aside, regardless of whether it is set aside before or after judgment is entered. In fact, though, the plurality really makes no attempt to explain the entry-of-judgment distinction in terms of either the at-the-time-of-senteneing-under-staté-law rule, or in terms of the inevitable-understate-law rule. Rather, the significance of the entry of judgment rests upon the assertion that a judgment is more certain than a jury verdict. The entry-of-judgment line, then, is really about relative degrees of certainty regarding parole ineligibility.17
If the question is not one in which state law controls (by looking to the defendant’s conclusively determined status either at the time of sentencing or inevitably thereafter), the question of Simmons’ applicability must be an issue of federal due process law. That is a proposition with which I agree entirely; indeed, Simmons itself makes that perfectly clear, as I discuss below. Before examining what Simmons’ *194due process requirements entail, however, it is important to understand the rationale behind Simmons: the need for capital sentencing juries to have accurate information about the defendant in the particular area of parole eligibility.
I — * H-i
We stated m Gregg v. Georgia, 428 U. S. 153 (1976):
“If an experienced trial judge, who daily faces the difficult task of imposing sentences, has a vital need for accurate information about a defendant and the crime he committed in order to be able to impose a rational sentence in the typical criminal case, then accurate sentencing information is an indispensable prerequisite to a reasoned determination of whether a defendant shall live or die by a jury of people who may never before have made a sentencing decision.” Id., at 190 (joint opinion of Stewart, Powell, and Stevens, JJ.).
This imperative is all the more critical when the jury must make a determination as to future dangerousness. “[A]ny sentencing authority must predict a convicted person’s probable future conduct when it engages in the process of determining what sentence to impose.... What is essential is that the jury have before it all possible relevant information about the individual defendant whose fate it must determine.” Jurek v. Texas, 428 U. S. 262, 274-276 (1976) (joint opinion of Stewart, Powell, and Stevens, JJ.). When it comes to issues such as future dangerousness and the possibility of parole, it is therefore vitally important that “the jury [have] accurate information of which both the defendant and his counsel are aware,” including “an accurate statement of a potential sentencing alternative.” California v. Ramos, 463 U. S. 992, 1004,1009 (1983).
This is not to say, of course, that the Constitution compels the States to tell the jury every single piece of information that may be relevant to its deliberations. See, e. g., id., at *1951010-1012. Indeed, in California v. Ramos, we held it ordinarily proper to “defe[r] to the State’s choice of substantive factors relevant to the penalty determination.” Id., at 1001. Notwithstanding the broad discretion recognized in Ramos, the latitude to-which the States are entitled is not unbounded; at times, it must give way to the demands of due process.
One such due process requirement is that a defendant must have an opportunity to rebut the State’s ease against him. Simmons, 512 U. S., at 175 (O’Connor, J., concurring in judgment). And “[w]hen the State seeks to show the defendant’s future dangerousness, ... the fact that he will never be released from prison will often be the only way that a violent criminal can successfully rebut the State’s ease.” Id., at 177 (O’Connor, J., concurring in judgment). Accordingly, “despite our general deference to state decisions regarding what the jury should be told about sentencing,... due process requires that the defendant be allowed [to bring his parole ineligibility to the jury’s attention] in cases in which the only available alternative sentence to death is life imprisonment without possibility of parole and the prosecution argues that the-defendant will pose a threat to society in the future.” Ibid.
The rationale for the Simmons exception to the general rule of Ramos is quite apparent. In Ramos, the defendant claimed that if the State were permitted to argue that the Governor could commute a sentence of life without parole, then due process entitled him to tell the jury that the Governor could commute a death sentence as well. We rejected that argument, however, holding that the information the defendant sought to introduce “would not ‘balance’ the impact” of telling the jury that the Governor could commute a sentence of life without parole. 463 U. S., at 1011. Nor would it make the jury “any less inclined to vote for the death penalty upon learning” that information. Ibid. Nor, finally, were we persuaded that it would “impermissibly impe[l] the *196jury toward voting for the death sentence” if the jury were told that a life without parole sentence could be commuted, but were not told that a death penalty could be commuted as well. Id., at 1012.
Each of these factors, however, points in precisely the opposite direction when it comes to information about a defendant’s parole ineligibility. If the State argues that the defendant will be a future danger to society, it quite plainly rebuts that argument to point out that the defendant — because of his parole ineligibility — -will never be a part of society again. Simmons, 512 U. S., at 177 (O’Connor, J., concurring in judgment) (“[TJhe fact that he will never be released from prison will often ... rebut the State’s case”). And unlike Ramos, if the jury is informed of a defendant’s parole ineligibility, it is “less inclined to vote for the death penalty upon learning” that fact. Conversely, permitting the State to argue the defendant’s future dangerousness, while simultaneously precluding the defendant from arguing his parole ineligibility, does tend to “impe[l] the jury toward voting for the death sentence.” Despite the plurality’s unsupported remark that “[ejvidence of potential parole ineligibility is of uncertain materiality,” ante, at 170, all of the available data demonstrate to the contrary.
How long a defendant will remain in jail is a critical factor for juries. One study, for example, indicates that 79% of Virginia residents consider the number of years that a defendant might actually serve before being paroled to be an “‘important consideration when choosing between life imprisonment and the death penalty.’”18 A similar study reveals that 76.5% of potential jurors think it is “extremely important”. or “very important” to know that information when deciding between life imprisonment and the death pen*197alty.19 Likewise, two-thirds of the respondents in another survey stated that they would be more likely to give a life sentence instead of death if they knew the defendant had to serve at least 25 years in prison before being parole eligible.20 General public support for the death penalty also plummets when the survey subjects are given the alternative of life without parole.21 Indeed, parole ineligibility information is so important that 62.3% of potential Virginia jurors would actually disregard a judge’s instructions not to consider parole eligibility when determining the defendant’s sentence.22
At the same time, the recent development of parole ineligibility statutes results in confusion and misperception, such that “common sense tells us that many jurors might not know whether a life sentence carries with it the possibility *198of parole.” Simmons, 512 U. S., at 177-178 (O’Connor, J., concurring in judgment). The statistical data bear this out. One study of potential Virginia jurors asked: “‘If a person is sentenced to life imprisonment for intentional murder during an armed robbery, how many years on the average do you think that the person would actually serve before being released on parole?”’ The most frequent response was 10 years.23 Another potential-juror survey put the average response at just over eight years.24 And more than 70% of potential jurors think that a person sentenced to life in prison for murder can be released at some point in the future.25
Given this data, it is not surprising that one study concluded: “[Jjurors assessing dangerousness attach great weight to the defendant’s expected sentence if a death sentence is not imposed. Most importantly, jurors who believe the alternative to death is a relatively short time in prison tend to sentence to death. Jurors who believe the alternative treatment is longer tend to sentence to life.”26 Consequently, every reason why the Governor’s commutation power at issue in Ramos was not required to be put before the jury leads to precisely the opposite conclusion when it comes to the issue of parole ineligibility. That is exactly why Simmons is an exception to the normally operative rule of deference established in Ramos.27
The plurality — focusing exclusively on one of the many sources cited — criticizes at length (ante, at 172-178) these “so-called scientific conclusions” that merely confirm what *199every sentencing judge surely knows — that how soon the defendant may actually be released from prison is highly relevant to the sentencing decision. The plurality’s criticism yet again underscores the formalistic character of its analysis of the life-or-death issue presented by this case. In exercising the judicial function, there are times when judgment is far more important than technical symmetry.28
IV
The Virginia Supreme Court held that whether Simmons applies is a question whose answer is entirely controlled by the operation of state law. See supra, at 184-185. This understanding was adopted by the plurality as well, at least as it originally stated the holding of Simmons. See supra, at 185. But as explained above, the Virginia court’s view, as well as the plurality’s original stance, simply cannot be reconciled with Simmons itself. That might explain why the plurality ultimately abandons that view, instead relying *200on an assessment of how probable it is that the defendant will be found parole ineligible — or, as the plurality might put it, what is “more certain” under state law.
The plurality is correct to reject the 'Virginia Supreme Court’s holding that state law entirely controls the applicability of Simmons. Simmons announced a rule of due process, not state law. 512 U. S., at 156 (plurality opinion); id., at 177 (O’Connor, J., concurring in judgment). This is not to say that the federal due process right in Simmons does not make reference to state law, for surely it does; the very reason why Simmons is an exception to Ramos is because of the consequences of parole ineligibility under state law. But that is not the same thing as saying that the precise, technical operation of state law entirely controls its applicability.
Simmons itself makes this perfectly clear. In that case South Carolina argued that “because future exigencies such as legislative reform, commutation, clemency, and escape might allow [Simmons] to be released into society, [Simmons] was not entitled to inform the jury that he is parole ineligible.” 512 U. S., at 166, and n. 6 (plurality opinion). Indeed, as noted earlier, it argued that Simmons was not, technically, parole ineligible at the time of sentencing because the state parole board had not yet made its determination. See supra, at 186-187.
Yet the plurality opinion rejected outright the argument that “hypothetical future developments” control the issue, finding that South Carolina’s argument about state law, while “technically . . . true,” and “legally accurate,” had “little force.” Simmons, 512 U. S., at 166, and n. 6.29 In other words, the due process standard of Simmons was not eon-*201trolled entirely by the technical minutiae of state law, even though it looked at state law for determining when the right to rebut the State’s argument was triggered.
It makes perfect sense for Simmons’ due process right to make reference to, yet not be wholly controlled by, state law. On the one hand, Simmons is a limited exception to Ramos, and as such it is confined to where the defendant will be parole ineligible — hence the reference to state law. On the other hand, Simmons is a constitutional requirement imposed on the States. If its applicability turned entirely on a defendant’s technical status under state law at the time of sentencing, the constitutional requirement would be easily evaded by the artful crafting of a state statute. For example, if Virginia can define “conviction” to require an entry of judgment, it could just as easily define “conviction” to require that all final appeals be exhausted, or that all state and federal habeas options be foreclosed. And by delaying when the defendant’s convictions count as strikes for parole ineligibility purposes until some point in time well after the capital murder sentencing phase, the State could convert the Simmons requirement into an opt-in constitutional rule.30
Simmons’ applicability is therefore a question of federal law, and that case makes clear that the federal standard essentially disregards future hypothetical possibilities even if they might make the defendant parole eligible at some *202point.31 The question in this case, then, boils down to whether the plurality’s line between entry of judgment and a verdict is a demarcation of Simmons’ applicability that is (1) consistent with Simmons; (2) a realistic and accurate assessment of the probabilities; and (3) a workable, clear rule. I believe the plurality fails on each score.
It is important to emphasize the precise basis for the uncertainty the plurality perceives. The plurality limits the relevant uncertainty to things known before the time of sentencing. Events developing the day after sentencing, which might lend uncertainty to a defendant’s eventual parole ineligibility do not make Simmons inapplicable, the plurality says. Ante, at 176. What I understand the plurality to be concerned about is whether the facts, as known at or prior to sentencing, east any doubt on whether, after sentencing, the defendant will become parole ineligible. Even if nothing definitive has happened yet by the time of sentencing, the facts as known at that time might well give rise to uncertainty as to the defendant’s parole ineligibility.
The question, then, is what were the facts as known at the time of Ramdass’ sentencing that might cast doubt on whether he would be found parole ineligible after sentencing. The facts to which the plurality points are, first, that judgment had not yet been entered on the verdict, and second, that the verdict could have been set aside if Ramdass had *203filed a motion to set aside the verdict under Rule 3A:15(b) and the trial court had found that motion meritorious. But no motion to set aside the verdict had been filed or was pending; no legal basis for granting such a motion had (or has) ever been identified; and there was not the slightest indication from the Domino’s Pizza robbery trial court that such a motion would have been found meritorious if it had been filed. In short, the plurality finds constitutionally significant uncertainty in the hypothetical possibility that a motion, if it had been filed, might have identified a trial error and the court possibly could have found the claim meritorious. The mere availability of a procedure for setting aside a verdict that is necessary for the defendant’s parole ineligibility is enough, the plurality says, to make Simmons inapplicable.
Frankly, I do not see how Simmons can be found inapplicable on the basis of such a “hypothetical future develop-men[t].” 512 U. S., at 166 (plurality opinion). The plurality offers no evidence whatsoever that this possibility — an “if only” wrapped in a “might have” inside of a “possibly so”— is at all more likely to occur than the “hypothetical future developments” that Simmons itself refused to countenance. Why is that possibility of setting aside the verdict any more likely than the fanciful scenarios dismissed in Simmons? Why is the certainty diminished merely because the trial judge has not yet entered judgment, when that fact has no bearing on whether a Rule 8A:15(b) motion will be granted? The plurality never tells us, for it simply declares, without support, elaboration, or explanation, that a verdict is more uncertain than a judgment is. See supra, at 192-193, and n. 17. The only reason it suggests for why the verdict here was uncertain is rather remarkable — that Ramdass himself said so. That is, the plurality relies upon the fact that a convicted murderer with minimal education and a history of drug experimentation including PCP and cocaine, App. 49, said “I don’t know” when asked if he could ever be released *204from prison. Ante, at 177. This evidence is thinner than gossamer.32
What’s more, the plurality’s assessment of certainties is internally inconsistent. As explained earlier, the standard for setting aside a verdict post-trial is the same regardless of whether judgment has been entered. Accordingly, if the verdict was uncertain in the Domino’s Pizza ease, that was also true for the Pizza Hut conviction. At the time of the sentencing hearing in the capital murder case, the deadline for filing a motion under the Rule had not expired for either the Domino’s Pizza verdict or the Pizza Hut conviction. (The time for filing a motion for the Pizza Hut conviction expired on February 12, 21 days after judgment had been entered on that verdict. This was 13 days after the sentencing phase in the capital murder case ended.) Because there was a possibility that the Pizza Hut conviction could have been set aside before judgment was entered on the Domino’s Pizza verdict (and therefore before Ramdass technically became parole ineligible), the certainty of the verdict was just as much in doubt for that conviction. The plurality, however, finds the Domino’s Pizza verdict uncertain yet casts no doubt on the Pizza Hut conviction. How can this possibly be consistent? The plurality never says.
Finally, the plurality’s approach is entirely boundless. If the kind of “hypothetical future development]” at issue hei’e is sufficient to make Simmons inapplicable, would it *205be sufficient if, rather than having the possibility of a recent conviction being set aside by post-trial motion, an old prior conviction could be set aside on appeal before judgment had been entered on the Domino’s Pizza verdict? Or under a State’s postconvietion habeas procedure? More to the point, if the mere availability of a post-trial procedure to set aside the verdict is enough, is the same true as well for the mere availability of an appeal or state habeas review, so long as the time had not expired for either? Old convictions necessary for a defendant’s parole ineligibility can be set aside under these procedures as well. And under each procedure those prior convictions could potentially be set aside at the crucial moment.33
It is easy, in this ease, to be distracted by the lack of an entry of judgment and the reeentness of Ramdass’ prior convictions. As the above examples demonstrate, however, these facts tend to detract from, rather than elucidate, the relevant issue. If Simmons is inapplicable because at least one of the defendant’s prior convictions could be set aside before sentencing (or before the third strike becomes final, or before whatever time the plurality-might think is the crucial moment), then it should not matter, under that reasoning, whether it is set aside by post-trial motion, on appeal, or through state (or federal) postconviction relief. What’s more, the plurality’s reasoning would hold true so long as these procedures are simply available. Accordingly, it would not matter whether a defendant’s prior strikes were a day old, a year old, or 100 years old. Nor would it matter that judgment had been entered on those prior convictions. So long as such procedures for setting aside old convictions *206exist and remain technically available prior to a defendant’s capital murder sentencing phase, the defendant’s eventual parole ineligibility is just as uncertain at the crucial moment.
The plurality, however, never addresses any of this, but surely its holding today is an invitation to such possibilities. Indeed, if these possibilities make Simmons inapplicable, does this not invite the very same circumvention of Simmons that would result if the rule turned entirely on state law (see swpra, at 201), by allowing a State to render all prior convictions uncertain simply by holding open some theoretical possibility for postconviction relief at all times? Given that appeals and various forms of posteonviction relief undermine the certainty of a verdict or a “conviction” every bit as much as does a procedure like Rule 8A:15(b)— indeed, probably more so — the plurality’s reasoning either draws an arbitrary line between these types of procedures, or it accepts that all of these possibilities make Simmons inapplicable, in which case that due process right is eviscerated entirely.34 It is abundantly clear that the proclaimed “workable” rule the plurality claims to be following is an illusion. Ante, at 166.
No such arbitrary line-drawing is at all necessary to decide this case. It is entirely sufficient simply to hold that Virginia has offered not one reason for doubting that judgment would be entered on the Domino’s Pizza robbery verdict or for doubting .Ramdass’ eventual parole ineligibility. Certainly it has offered no reason for thinking that the possibil*207ity of setting aside the Domino’s Pizza robbery verdict is at all more likely than the hypothetical future developments rejected in Simmons. This ease thus falls squarely within Simmons.
Though it is unnecessary to decide it here, a guilty verdict is the proper line. A guilty verdict against the defendant is a natural breaking point in the uncertainties inherent in the trial process. Before that time, the burden is on the State to prove the accused’s guilt beyond a reasonable doubt. A guilty verdict, however, means that the defendant’s presumption of innocence — with all of its attendant trial safeguards — has been overcome. The verdict resolves the central question of the general issue of guilt. It marks the most significant point of the adversary proceeding, and reflects a fundamental shift in the probabilities regarding the defendant’s fate. For that reason, it is the proper point at which a line separating the hypothetical from the probable should be drawn. Moreover, because the State itself can use the defendant’s prior crimes to argue future dangerousness after a jury has rendered a verdict — as Virginia did here, see supra, at 182-183, and n. 2 — that is also the point at which the defendant’s Simmons right should attach.
V
In Williams v. Taylor, 529 U. S. 362, 405 (2000) (opinion of O’Connor, J.), we stated the standard for granting habeas relief under 28 U. S. C. § 2254(d)(1): “A state-eourt decision will certainly be contrary to our clearly established precedent if the state court applies a rule that contradicts the governing law set forth in our eases.” As I have explained, the Virginia Supreme Court applied Simmons as if (a) its applicability was controlled entirely by state law and (b) the defendant’s parole ineligibility is determined at the exact moment when the sentencing phase occurs. See supra, at 184-185. But state law does not control Simmons’ applicability, nor does the due process right turn on whether the defend*208ant has already been found parole Ineligible at the exact moment of sentencing. Simmons itself makes this entirely clear. Both aspects of the Virginia Supreme Court’s holding, then, applied a “rule that contradicts the governing law set forth in” Simmons.
We also held in Williams that “[a] state-court decision will also be contrary to this Court’s clearly established precedent if the state court confronts a set of facts that are materially indistinguishable from a decision of this Court and nevertheless arrives at a result different from our precedent.” 529 U. S., at 406. The Virginia Supreme Court’s decision was also contrary to Simmons in this respect. Because the “hypothetical future developments” rejected in Simmons are materially indistinguishable from the future possibility here, the Virginia court’s decision is contrary to Simmons.
Even assuming the correct rule had been applied, the Virginia Supreme Court’s decision would be an “unreasonable application” of Simmons. That court held that the Pizza Hut conviction would count as a strike, but not the Domino’s Pizza robbery verdict. The only distinction is the lack of an entry of judgment, and the only reason that matters is because the verdict may be set aside by a post-trial motion. But that possibility remains identical for both crimes. To disregard one of those hypothetical possibilities but not the other based on a state-law distinction that has absolutely no relevance to the probability that the verdict will be set aside is an unreasonable application of Simmons.35
*209< WH
Nothing in the above arguments should distract us from the fact that this is a simple case. The question turns on whether the hypothetical possibility that the trial judge might fail to sign a piece of paper entering judgment on a guilty verdict should mean that the defendant is precluded from arguing his parole ineligibility to the jury. We should also not be distracted by the plurality’s red herring — the possibility of setting aside the verdict by a post-trial motion. Not only is that possibility indistinguishable from the non-exhaustive list of hypothetical future possibilities we dismissed in Simmons, but it also fails to distinguish this case from the many other possibilities that are part of the state criminal justice system, and fails to distinguish Ramdass’ convictions from each other.
The plurality’s convoluted understanding of Simmons and its diverse implications necessitate a fair amount of disentangling of its argument. But, once again, this should not divert us from the plain reality of this ease. Juries want to know about parole ineligibility. We know how important *210it is to their life-and-death decisionmaking. We know how misinformed they are likely to be if we do not give them this information. We know Simmons has worked,36 and we know the States have wholeheartedly embraced it.37
Moreover, we know this jury thought the information was critical; we know this jury misunderstood what a “life” sentence meant; we know this jury would have recommended life instead of death if it had known that Ramdass was parole ineligible; and we know this jury did not get a clear answer to its question. We also know that Virginia entrusts to the jury the solemn duty of recommending life or death for the defendant. Why does the Court insist that the Constitution permits the wool to be pulled over their eyes?
I respectfully dissent.
The prosecution’s opening argument began by recounting Ramdass’ entire criminal history. App. 8-11. Eight of the nine witnesses the Commonwealth called did little more than relate the details of Ram-dass’ criminal past. Id., at 12-64. - The prosecution’s closing argument highlighted the connection between Ramdass’ crimes and his prior releases from prison. Id., at 80-82. In fact, it did so on several occasions. Id., at 9 (Ramdass “served time [for the 1988 strong arm robbery conviction] and was finally paroled in May of 1992”); id., at 46-47 (Ramdass “was *183released on mandatory parole” in 1992, shortly before his most recent crime spree began); id., at 51b-52 (describing Ramdass’ 1992 release on mandatory parole).
Id., at 57-59 (“On that next night, August 30th, you did a robbery of the Domino’s Pizza over in Alexandria? . . . Well, if the cab driver was shot in the head on August 30th and Domino’s Pizza was August 30th, you did them both the same day; didn’t you?”); id., at 81 (“August 30th, 1992, he robbed Domino’s Pizza at the point of a gun in Alexandria and he robbed Domino’s Pizza not long after he shot that Arlington cab driver through the head...”).
Of course, Simmons v. South Carolina, 512 U. S. 154 (1994), applies when the prosecution argues future dangerousness; it does not require the State to argue any particular past crime. My purpose in pointing out Virginia’s reliance on the Domino’s Pizza verdict is to underscore the unfairness of permitting Virginia to use it, while denying Ramdass the same use. The plurality’s repeated statement that Virginia brought up the crime in its cross-examination rather than its case in chief, ante, at 162, 170, 171, neither means Simmons is inapplicable nor mitigates the unfairness here. It only signals the formalism the plurality is prepared to endorse.
Once again, Simmons’ applicability does not at all turn on whether this kind of evidence exists. I point it out only to emphasize how real the Simmons concerns are here. The plurality complains, in essence, that the evidence came in the form of an uncontested proffer rather than as a sworn affidavit. Ante, at 163. Again, neither Simmons’ applicability nor the reality of the case is undercut by this quibble. The only thing that it proves is the plurality's penchant for formalism.
Though the plurality does not include the proviso in its initial statement of the rule in Simmons, it repeats this requirement no less than 20 times in its 20-page opinion. See ante, at 159 (“when the jury considered his case”), 161 (“at the time of the capital sentencing trial”), 163-164 (“at the time of the jury’s death penalty deliberations”), 164 (“when the jury was considering his sentence”), ibid, (“at the time of the sentencing trial”), ibid, (“at the time of his trial”), ibid, (“at the time of sentencing”), 165 (“at the time of the jury’s future dangerousness deliberations”), 166 (“when the jury deliberated his sentence”), ibid, (“at the time of the sentencing trial”), 167 (“when the jury considered his sentence”), ibid, (“at the time of sentencing”), ibid, (“at the time of his sentencing trial”), ibid, (same), 168 (“at the time of the sentencing trial”), ibid, (same), ibid, (“at the time of his sentencing trial”), 169 (“at the time of his trial”), 171 (“at the time of his trial”), 176 (“at the time of trial”).
That statute read in part: “The board must not grant parole nor is parole authorized, to any prisoner serving a sentence for a second or subsequent conviction, following a separate sentencing for a prior conviction, for violent crimes as defined in Section 16-1-60. Provided that where more than one included offense shall be committed within a one-day period or pursuant to one continuous course of conduct, such multiple offenses must be treated for purposes of this section as one offense.”
See, e. g., State v. McKay, 300 S. C. 113, 115, 386 S. E. 2d 623,623-624 (1989).
It is true, as the plurality points out, ante, at 167, that in Simmons the defendant did have an entry of judgment. But, under the plurality’s reasoning, the issue is whether the defendant is parole ineligible at the time of sentencing, not why he is or is not ineligible. Thus, whether the defendant is parole eligible at that time because he has no entry of judgment or because the parole board has not yet met is hardly relevant. It is a distinction, but not a material one.
“First and foremost, at the time of the trial, no state agency had ever determined that Simmons was going to be serving a sentence of life without the possibility of parole, despite the fact that he had earlier pled guilty and been sentenced to a violent crime prior to this trial. The importance of that distinction is that the power to make that determination did not rest with the judiciary, but was solely vested in an executive branch agency, the South Carolina Board of Probation, Parole and Pardon Services.” Brief for Respondent in Simmons v. South Carolina, O. T. 1993, No. 92-9059, p. 95 (emphasis added).
The plurality also complains that “a state court [need not] glean information from the record” in Simmons. Ante, at 168. That is true, but it is equally true that a state court cannot pretend that a fact creates a material distinction simply because it was not expressly raised and rejected by this Court. Moreover, it is evident in the opinion itself that Simmons’ parole-ineligibility status had not been definitively and legally determined yet at the time of sentencing. See n. 8, infra.
The plurality contends that in Simmons “the defendant had ‘conclusively established’ his parole ineligibility at the time of sentencing.” Ante, at 167 (quoting Simmons, 512 U. S., at 158 (plurality opinion)); see also ante, at 171. What Simmons in fact said was that no one questioned that the defendant had all the facts necessary to be found ineligible at some future date. It does not indicate that a legal determination of the defendant’s parole ineligibility had already been definitively made by the parole board. This is dear in the plurality’s dtation of the South Carolina parole statute, under which a defendant’s parole status is determined by the parole board at a later date. See supra, at 186. This is also dear from the fact that the plurality relied upon the testimony of the parole board’s attorneys, 512 U. S., at 158-159, demonstrating the plurality’s recognition that it was the parole board that would ultimately determine Simmons’ parole eligibility. Furthermore, the plurality’s statement that Simmons was “in fact ineligible,” id., at 158 (emphasis added), as opposed to “legally” ineligible or ineligible “as a matter of law,” dearly distinguished between the facts as known at that time (which indicated how Simmons’ status would, in all likelihood, ultimately be determined), and the legal determination of status (which would be formally determined at *188a later date). Finally, if Simmons’ parole ineligibility had been legally and conclusively resolved by the time of his trial, there would have been no need for the plurality to discuss (and reject) possibilities that might have undermined Simmons’ eventual finding of parole ineligibility. See infra, at 201-203.
The Simmons plurality did say that “an instruction informing the jury that petitioner is ineligible for parole is legally accurate.” 512 U. S., at 166; ante, at 166-167. But in the very next sentence the plurality wrote: “Certainly, such an instruction is more accurate than no instruction at all.” 512 U. S., at 166 (emphasis added). This made it clear that “accuracy,” in the sense used there, is a relative term, not an absolute conclusive determination of legal status.
At the time of Ramdass’ trial, Rule 3A:15(b) read:
“(b) Motion to Set Aside Verdict. — If the jury returns a verdict of guilty, the court may, on motion of the accused made not later than 21 days after entry of a final order, set aside the verdict for error committed during the trial or if the evidence is insufficient as a matter of law to sustain a conviction.”
South Carolina Rule of Criminal Procedure 29(b) (1999) reads, in relevant part: “A motion for a new trial based on after-discovered evidence must be made within a reasonable period of time after the discovery of the evidence.”
It is true, of course, that a motion for a new trial under South Carolina’s rule must be predicated on the discovery of new evidence, but that does not meaningfully distinguish its rule from Virginia’s rule, under which a verdict can be set aside only for trial error or insufficient evidence.
The plurality says that because Simmons pleaded guilty to his prior crime, he was foreclosed from filing a motion under South Carolina’s rule. Ante, at 167. For this proposition, the plurality cites Whetsell v. State, 276 S. C. 295, 277 S. E. 2d 891 (1981). This is just flat wrong. See Johnson v. Catoe, 336 S. C. 354,358-359,520 S. E. 2d 617,619 (1999) <“Whetsell does not stand for the proposition that a defendant who admits his guilt is barred from collaterally attacking his conviction. Whetsell stands only for the narrow proposition that a [postconviction relief] applicant who has pled guilty on advice of counsel cannot satisfy the prejudice prong on collateral attack if he states he would have pled guilty in any event”).
Dowell v. Commonwealth, 12 Va. App. 1145, 408 S. E. 2d 263 (1991); Davis v. Commonwealth, No. 2960-98-2,2000 WL 135148 (Va. App., Feb. 8, 2000) (unpublished).
See id., at *4, n. 1.
Floyd v. Commonwealth, 219 Va. 575, 577, 249 S. E. 2d 171, 172 (1978) (“Overruling Floyd’s motions to set aside the verdicts ..., the trial court entered judgments on the verdicts”); Johnson v. Commonwealth, 20 Va. App. 547, 552,458 S. E. 2d 599, 601 (1995) (“At Johnson’s sentencing hearing, defense counsel made a motion to set aside the verdict.... The trial judge denied the motion”); Walker v. Commonwealth, 4 Va. App. 286,291, 356 S. E. 2d 853, 856 (1987) (“After the jury was discharged, defendant *191moved the court to set aside the verdict; the court denied the motion”); Carter v. Commonwealth, 10 Va. App. 507,509,393 S. E. 2d 639,640 (1990) (“Carter . . . appeals from judgments of the Circuit Court of Loudoun County... which ... denied his post-trial motions for a new trial”).
Payne v. Commonwealth, 220 Va. 601, 602-603, 260 S. E. 2d 247, 248 (1979).
Dowell does say that a verdict without a judgment is not reliable “for impeachment purposes,” 12 Va. App., at 1149,408 S. E. 2d, at 265, but this is a far cry from saying the verdict is itself unreliable. What the three cases actually address is the question whether a verdiet is a “conviction” under state law; they say that it depends on the context, answering in the negative in two cases, and in the affirmative in a third.
The plurality also cites two intermediate state-court cases making passing reference to a trial court’s granting of a post-trial motion, though neither case mentions Rule 3A:15(b). See Gorham v. Commonwealth, 15 Va. App. 673, 426 S. E. 2d 493 (1993); Cullen v. Commonwealth, 13 Va. App. 182,409 S. E. 2d 487 (1991).' But a mere two cases among all the criminal *192eases in Virginia surely demonstrates that setting aside a verdict by post-trial motion is a rarity; if those two instances make the verdict uncertain, then one might as well cite the solitary case in which the Governor granted a pardon after the verdict but before the entry of judgment. See Blair v. Commonwealth, 66 Va. 850 (1874).
Though the plurality insists that judgment “is the usual measure for finality,” ante, at 176, its own opinion reveals that it does not mean “finality” in any absolute sense. Rather, it concedes that while a “jury verdict [isl uncertain,” ante, at 175, “even a judgment” is “uncertain” too, because of "the availability of postjudgment relief,” ante, at 176. What it means, then — though it is not particularly candid about it — is that a judgment is more certain than a verdict. Put differently, the plurality thinks a judgment is more enduring, in that there is a greater probability that a verdict will survive a motion to set it aside if there has already been an entry of judgment.
It is dear that the significance of the entry of judgment for the plurality must be based on that belief The significance cannot be that without the entry of judgment the defendant is not ineligible for parole at the exact moment of sentencing; as explained above, that fact is not dispositive. See supra, at 185-187. Nor can its significance be that without the entry of judgment, his parole status is not inevitable. As also explained above, the entry of judgment has no significance insofar as inevitability is concerned. See supra, at 188-192 and this page.
See Note, The Meaning of “Life” for Virginia Jurors and Its Effect on Reliability in Capital Sentencing, 75 Va. L. Rev. 1605, 1624, and n. 102 (1989) (citing study by National Legal Research Group).
Hughes, Informing South Carolina Capital Juries About Parole, 44 S. C. L. Rev. 383,409-410 (1993) (citing 1991 study by Univ. of South Carolina’s Institute for Public Affairs); see also Simmons, 512 U. S., at 159 (plurality opinion) (discussing this study).
Paduano & Smith, Deathly Errors: Juror Misperceptions Concerning Parole in the Imposition of the Death Penalty, 18 Colum. Human Rights L. Rev. 211,223 (1987).
See, e. g., Rising Doubts on Death Penalty, USA Today, Dec. 22, 1999, p. 17A (nationwide 1999 Gallup Poll finds 71% support death penalty; 56% support death penalty when life without parole is offered as an option); Finn, Given Choice, Va. Juries Vote for Life, Washington Post, Feb. 3, 1997, pp. Al, A6 (“According to a poll conducted for the Death Penalty Information Center, which opposes capital punishment, support for the death penalty nationwide falls from 77 percent to 41 percent if the alternative is life without parole accompanied by restitution”); Heyser, Death Penalty on the Rise in Virginia, Roanoke Times, Aug. 31, 1998, p. C3 (reporting study by Virginia Tech’s Center for Survey Research, finding that 79% of Virginians “strongly” or "somewhat” support the death penalty, a figure that drops to 57% when respondents are given the alternative of life without parole for 25 years plus restitution); Armstrong & Mills, Death Penalty Support Erodes, Many Back Life Term as an Alternative, Chicago Tribune, Mar. 7,2000, p. 1 (58% of Illinois registered voters support death penalty; only 43% favor death when given option of life without parole).
See Note, 75 Va. L. Rev., at 1624-1625, and n. 103.
See id., at 1624, and n. 101.
See Paduano & Smith, 18 Colum. Human Rights L. Rev., at 223, n. 34.
See Hughes, 44 S. C. L. Rev., at 408; see also Finn, Washington Post, at A6 (“[O]nly 4 percent of Americans believe that convicted murderers will spend the rest of their days in prison”).
See Eisenberg & Wells, Deadly Confusion: Juror Instructions in Capital Cases, 79 Cornell L. Rev. 1, 7 (1993).
See Simmons, 512 U. S., at 159, 170, n. 9 (plurality opinion) (discussing above data).
As for the specific criticisms, the plurality first complains that such surveys are inadmissible as evidence. The question, though, is not whether the statistical studies are admissible evidence, but whether they are relevant facts assisting in our determination of the proper scope of the Simmons due process right. Surely they are. In any event, Ramdass did raise such studies at his sentencing hearing. See App. 95-96. Virginia had its chance to object, but opted not to do so. It is far too late in the day to complain about it now. (Simmons, incidentally, also introduced similar evidence in his trial without objection. See 512 U. S., at 159 (plurality opinion).)
Next, the plurality says that one of the studies I cited focused only on Georgia jurors, as if Georgians have some unique preference for life without parole. In any event, the studies focusing on Virginia jurors yield the same results. See nn. 18,21, supra. Finally, the plurality questions the objectivity of one particular study. Even if the plurality were justified in that criticism, it surely has no basis for questioning the many other sources cited. See n. 19, supra (Univ. of South Carolina’s Institute for Public Affairs), n. 21 (Gallup Poll and Virginia Tech’s Center for Survey Research), n. 26 (study by Associate Professor of Statistics, Dept, of Economic and Social Statistics, Cornell Univ.).
While Justice O’Connor’s concurring opinion did not make direct reference to those hypothetical possibilities, South Carolina's brief and the plurality’s opinion put the issue squarely before the Court. If those hypothetical had made a difference, the outcome of the case for the concurring opinion would have been precisely the opposite of what it was.
This is true even if one accepts the premise that Simmons requires us to presume that the most recent conviction will ultimately count as a strike regardless of what could happen under state law after the sentencing hearing. (The Virginia Supreme Court apparently adopted that view, which explains why that court counted the capital murder verdict as a strike at the time of the sentencing hearing, even though judgment had not yet been entered on the verdict. See supra, at 184.) Even accepting that premise, delaying the determination of parole ineligibility status until after the sentencing hearing would still mean that the defendant’s other prior convictions would not count as strikes until well after the capital murder sentencing phase.
The plurality’s claim, ante, at 169, that Ramdass seeks an extension of Simmons is therefore unfounded. And its criticism that “[pletitioner’s proposed rule would require courts to evaluate the probability of future events” ignores the fact that Simmons itself did the very same thing. Ante, at 169. The irony of that comment, moreover, is that it criticizes the rule for requiring an assessment of the future on the ground that such an inquiry is inherently speculative. Yet speculation about the future is precisely what is required when the jury is asked to assess a defendant’s future dangerousness. The speculation, however, becomes reasoned prediction rather than arbitrary guesswork only when the jury is permitted to learn of the defendant’s future parole status. See supra, at 194-199. Unfortunately, that was not the case here.
The plurality also attempts to distinguish the hypothetical in Simmons from those in Ramdass’ case by pointing out that the former hypo-thetieals, if they happened, would do so after sentencing. Ante, at 168— 169. But the entire point of the hypothetical is not whether they could occur before sentencing, but whether they could occur before the defendant was technically declared parole ineligible. In Simmons, that was true right up until the parole board made its determination. Simply because the nuances of state law may create an opportunity for undermining parole ineligibility earlier on does not make the possibility any less hypothetical or undermine the ineligibility any less; the same principle is at work either way.
It is true that these old convictions — like the Pizza Hut conviction— have had an entry of judgment and thus would count as strikes. But under state law, a defendant must have three strikes at the same time to be parole ineligible. If a strike were set aside before the defendant has all three, he is just as much parole eligible as he would be if judgment had never been entered on the verdict.
The plurality says “[t]he Commonwealth is entitled to some deference, in the context of its own parole laws, in determining the best reference point for making the ineligibility determination.” Ante, at 170; see also ante, at 176 (“States may take different approaches and we see no support for a rule that would require a State to declare a conviction final for purposes of a three-strikes statute once a verdict has been rendered”). But the questions here are whether the federal due process standard must abide by every state-law distinction, and if not, is abiding by the entry-of-judgment distinction arbitrary, in light of the fact that that distinction has absolutely no bearing on whether the verdict will be set aside?
Three remaining points should be addressed. First, Teague v. Lane, 489 U. S. 288 (1989), does not bar relief. Teague’s antiretroactivity doctrine is irrelevant here, as Simmons was decided before Ramdass' conviction became final. See 187 F. 3d 396, 404, n. 3 (CA4 1999) (case below). Nor is Teagues bar of applying “new rules” on federal habeas review any barrier; because Ramdass’ ease falls squarely within Simmons, that case controls entirely, and no new rule is necessary.
The second point concerns the plurality’s suggestion that Ramdass might have waived his Simmons claim. See ante, at 162-163,177. It is not neeessary to discuss the issue at length. It suffices to note that this *209is precisely the argument that Virginia raised on remand to the Virginia Supreme Court. That court was not persuaded by the argument, nor was any court during the entire state and federal habeas proceedings. See, e. g., App. 219, 225-226,281-284 (Magistrate’s Report) (discussing its own and other courts’ rejection of waiver argument); 187 F. 3d, at 402 (case below) (same). It is therefore not surprising that Virginia failed to argue waiver in its brief in opposition and arguments not raised therein are themselves normally deemed waived. See Caterpillar Inc. v. Lewis, 519 U. S. 61, 75, n. 13 (1996).
Finally, that Ramdass’ counsel argued that he would go to jail “for the rest of Ms life” does not at all satisfy Simmons’ requirement. Ante, at 161. The entire point of Simmons is that the jury will often misunderstand what it means to sentence a defendant to “life.” Consequently, that Ramdass was able to tell the jury he would get “life” simply does not help unless he is also permitted to tell the jury that life means life without the possibility of parole. Indeed, the very fact that the jury’s question came after counsel made this argument demonstrates that the jury was uncertain about what that statement meant.
See, e. g., Finn, Washington Post, at A1 (recounting how, after Virginia adopted life without parole alternative in 1995, and after Simmons, “[t]he number of people given the death sentence in Virginia has plummeted,” and describing “[sjimilar declines ... in Georgia and Indiana” as well as in Maryland).
See Yarbrough v. Commonwealth, 258 Va. 847, 519 S. E. 2d 602 (1999) (extending Simmons to apply even when State does not argue future dangerousness); ante, at 178.