concurring in the judgment.
The plurality and dissent agree on two fundamental propositions: that there is some age below which a juvenile’s crimes can never be constitutionally punished by death, and that our precedents require us to locate this age in light of the “ ‘evolving standards of decency that mark the progress of a maturing society.’ ” See ante, at 821 (quoting Trop v. Dulles, 356 U. S. 86, 101 (1958) (opinion of Warren, C. J.)); ante, at 827-829; post, at 864-865, 872. See also, e. g., McCleskey v. Kemp, 481 U. S. 279, 300 (1987). I accept both principles. The disagreements between the plurality and the dissent rest on their different evaluations of the evidence available to us about the relevant social consensus. Although I believe that a national consensus forbidding the execution of any person *849for a crime committed before the age of 16 very likely does exist, I am reluctant to adopt this conclusion as a matter of constitutional law without better evidence than we now possess. Because I conclude that the sentence in this case can and should be set aside on narrower grounds than those adopted by the plurality, and because the grounds on which I rest should allow us to face the more general question when better evidence is available, I concur only in the judgment of the Court.
I
Both the plurality and the dissent look initially to the decisions of American legislatures for signs of a national consensus about the minimum age at which a juvenile’s crimes may lead to capital punishment. Although I agree with the dissent’s contention, post, at 865, that these decisions should provide the most reliable signs of a society-wide consensus on this issue, I cannot agree with the dissent’s interpretation of the evidence.
The most salient statistic that bears on this case is that every single American legislature that has expressly set a minimum age for capital punishment has set that age at 16 or above. See ante, at 829, and n. 30. When one adds these 18 States to the 14 that have rejected capital punishment completely, see ante, at 826, and n. 25, it appears that almost two-thirds of the state legislatures have definitely concluded that no 15-year-old should be exposed to the threat of execution. See also ante, at 829, n. 29 (pointing out that an additional two States with death penalty statutes on their books seem to have abandoned capital punishment in practice). Where such a large majority of the state legislatures have unambiguously outlawed capital punishment for 15-year-olds, :and where no legislature in this country has affirmatively and unequivocally endorsed such a practice, strong counterevi-dence would be required to persuade me that a national consensus against this practice does not exist.
*850The dissent argues that it has found such counterevidence in the laws of the 19 States that authorize capital punishment without setting any statutory minimum age. If we could be sure that each of these 19 state legislatures had deliberately chosen to authorize capital punishment for crimes committed at the age of 15, one could hardly suppose that there is a settled national consensus opposing such a practice. In fact, however, the statistics relied on by the dissent may be quite misleading. When a legislature provides for some 15-year-olds to be processed through the adult criminal justice system, and capital punishment is available for adults in that jurisdiction, the death penalty becomes at least theoretically applicable to such defendants. This is how petitioner was rendered death eligible, and the same possibility appears to exist in 18 other States. See post, at 861-862; ante, at 828, n. 26. As the plurality points out, however, it does not necessarily follow that the legislatures in those jurisdictions have deliberately concluded that it would be appropriate to impose capital punishment on 15-year-olds (or on even younger defendants who may be tried as adults in some jurisdictions). See ante, at 826, n. 24.
There are many reasons, having nothing whatsoever to do with capital punishment, that might motivate a legislature to provide as a general matter for some 15-year-olds to be channeled into the adult criminal justice process. The length or conditions of confinement available in the juvenile system, for example, might be considered inappropriate for serious crimes or for some recidivists. Similarly, a state legislature might conclude that very dangerous individuals, whatever their age, should not be confined in the same facility with more vulnerable juvenile offenders. Such reasons, would suggest nothing about the appropriateness of capital punishment for 15-year-olds. The absence of any such implication is illustrated by the very States that the dissent cites as evidence of a trend toward lowering the age at which juveniles may be punished as adults. See post, at 867, and n. 3. New York, *851which recently adopted legislation allowing juveniles as young as 13 to be tried as adults, does not authorize capital punishment under any circumstances. In New Jersey, which now permits some 14-year-olds to be tried as adults, the minimum age for capital punishment is 18. In both cases, therefore, the decisions to lower the age at which some juveniles may be treated'as adults must have been based on reasons quite separate from the legislatures’ views about the minimum age at which a crime should render a juvenile eligible for the death penalty.
Nor have we been shown evidence that other legislatures directly considered the fact that the interaction between their capital punishment statutes and their juvenile offender statutes could in theory lead to executions for crimes committed before the age of 16. The- very real possibility that this result was not considered is illustrated by the recent federal legislation, cited by the dissent* which lowers to 15 the age at which a defendant may be tried as an adult. See post, at 865 (discussing Comprehensive Crime Control Act of 1984, Pub. L. 98-473, 98 Stat. 2149). Because a number of federal statutes have long provided for capital punishment, see post, at 866, n. 1, this legislation appears to imply that 15-year-olds may now be rendered death eligible under federal law. The dissent does not point to any legislative history suggesting that Congress considered this implication when it enacted the Comprehensive Crime Control Act. The apparent absence of such legislative history is especially striking in light of the fact that the United States has agreed by treaty to set a minimum age of 18 for capital punishment in certain circumstances. See Article 68 of the Geneva Convention Relative to the Protection of Civilian Persons in Time of War, August 12, 1949, [1955] 6 U. S. T. 3516, 3560, T. I. A. S. No. 3365 (rules pertaining to military occupation); ante, at 831, n. 34; see also ibid, (citing two other international agreements, signed but not ratified by the United States, prohibiting capital punishment for juveniles). Perhaps even more striking is *852the fact that the United States Senate recently passed a bill authorizing capital punishment for certain drug offenses, but prohibiting application of this penalty to persons below the age of 18 at the time of the crime. 134 Cong. Rec. 14117, 14118 (1988). Whatever other implications the ratification of Article 68 of the Geneva Convention may have, and whatever effects the Senate’s recent action may eventually have, both tend to undercut any assumption that the Comprehensive Crime Control Act signals a decision by Congress to authorize the death penalty for some 15-year-old felons.
Thus, there is no indication that any legislative body in this country has rendered a considered judgment approving the imposition of capital punishment on juveniles who were below the age of 16 at the time of the offense. It nonetheless is true, although I think the dissent has overstated its significance, that the Federal Government and 19 States have adopted statutes that appear to have the legal effect of rendering some of these juveniles death eligible. That fact is a real obstacle in the way of concluding that a national consensus forbids this practice. It is appropriate, therefore, to examine other evidence that might indicate whether or not these statutes are inconsistent with settled notions' of decency in our society.
In previous cases, we have examined execution statistics, as well as data about jury determinations, in an effort to discern whether the application of capital punishment to certain classes of defendants has been so aberrational that it can be considered unacceptable in our society. See, e. g., Coker v. Georgia, 433 U. S. 584, 592 (1977) (plurality opinion); Enmund v. Florida, 458 U. S. 782, 794-796 (1982); id., at 818-819 (O’Connor, J., dissenting). In this case, the plurality emphasizes that four decades have gone by since the last execution of a defendant who was younger than 16 at the time of the offense, and that only 5 out of 1,393 death sentences during a recent 5-year period involved such defendants. *853Ante, at 832-833. Like the statistics about the behavior of legislatures, these execution and sentencing statistics support the inference of a national consensus opposing the death penalty for 15-year-olds, but they are not dispositive.
A variety of factors, having little or nothing to do with any individual’s blameworthiness, may cause some groups in our population to commit capital crimes at a much lower rate than other groups. The statistics relied on by the plurality, moreover, do not indicate how many juries have been asked to impose the death penalty for crimes committed below the age of 16, or how many times prosecutors have exercised their discretion to refrain from seeking the death penalty in cases where the statutory prerequisites might have been proved. Without such data, raw execution and sentencing statistics cannot allow us reliably to infer that juries are or would be significantly more reluctant to impose the death penalty on 15-year-olds than on similarly situated older defendants.
Nor, finally, do I believe that this case can be resolved through the kind of disproportionality analysis employed in Part V of the plurality opinion. I agree that “proportionality requires a nexus between the punishment imposed and the defendant’s blameworthiness.” Enmund, supra, at 825 (O’Connor, J., dissenting); see also Tison v. Arizona, 481 U. S. 137 (1987). Granting the plurality’s other premise— that adolescents are generally less blameworthy than adults who commit similar crimes — it does not necessarily follow that all 15-year-olds are incapable of the moral culpability that would justify the imposition of capital punishment. Nor has the plurality educed evidence demonstrating that 15-year-olds as a class are inherently incapable of being deterred from major crimes by the prospect of the death penalty.
Legislatures recognize the relative immaturity of adolescents, and we have often permitted them to define age-based classes that take account of this qualitative difference between juveniles and adults. See, e. g., Hazelwood School *854District v. Kuhlmeier, 484 U. S. 260 (1988); Schall v. Martin, 467 U. S. 253 (1984); McKeiver v. Pennsylvania, 403 U. S. 528 (1971); Ginsberg v. New York, 390 U. S. 629 (1968). But compare Planned Parenthood of Central Missouri v. Danforth, 428 U. S. 52, 74-75 (1976) (unconstitutional for a legislature to presume that all minors are incapable of providing informed consent to abortion), and Bellotti v. Baird, 443 U. S. 622, 654 (1979) (Stevens, J., joined by Brennan, Marshall, and Blackmun, JJ., concurring in judgment) (same), with Akron v. Akron Center for Reproductive Health, Inc., 462 U. S. 416, 469, n. 12 (1983) (O’Con-nor, J., dissenting) (parental notification requirements may be constitutional). The special qualitative characteristics of juveniles that justify legislatures in treating them differently from adults for many other purposes are also relevant to Eighth Amendment proportionality analysis. These characteristics, however, vary widely among different individuals of the same age, and I would not substitute our inevitably subjective judgment about the best age at which to draw a line in the capital punishment context for the judgments of the Nation’s legislatures. Cf. Enmund, supra, at 826, and n. 42 (O’Connor, J., dissenting).
The history of the death penalty instructs that there is danger in inferring a settled societal consensus from statistics like those relied on in this case. In 1846, Michigan became the first State to abolish the death penalty for all crimes except treason, and Rhode Island soon thereafter became the first jurisdiction to abolish capital punishment completely. F. Zimring & G. Hawkins, Capital Punishment and the American Agenda 28 (1986). In succeeding decades, other American States continued the trend towards abolition, especially during the years just before and during World War I. Id., at 28-29. Later, and particularly after World War II, there ensued a steady and dramatic decline in executions — both in absolute terms and in relation to the number of homicides occurring in the country. W. Bowers, Legal Homicide *85526-28 (1984). In the 1950’s and 1960’s, more States abolished or radically restricted capital punishment, and executions ceased completely for several years beginning in 1968. H. Bedau, The Death Penalty in America 23, 25 (3d ed. 1982).
In 1972, when this Court heard arguments on the constitutionality of the death penalty, such statistics might have suggested that the practice had become a relic, implicitly rejected by a new societal consensus. Indeed, counsel urged the Court to conclude that “the number of cases in which the death penalty is imposed, as compared with the number of cases in which it is statutorily available, reflects a general revulsion toward the penalty that would lead to its repeal if only it were more generally and widely enforced.” Furman v. Georgia, 408 U. S. 238, 386 (1972) (Burger, C. J., dissenting). We now know that any inference of a societal consensus rejecting the death penalty would have been mistaken. But had this Court then declared the existence of such a consensus, and outlawed capital punishment, legislatures would very likely not have been able to revive it. The mistaken premise of the decision would have been frozen into constitutional law, making it difficult to refute and even more difficult to reject.
The step that the plurality would take today is much narrower in scope, but it could conceivably reflect an error similar to the one we were urged to make in Furman. The day may come when we must decide whether a legislature may deliberately and unequivocally resolve upon a policy authorizing capital punishment for crimes committed at the age of 15. In that event, we shall have to decide the Eighth Amendment issue that divides the plurality and the dissent in this case, and we shall have to evaluate the evidence of societal standards of decency that is available to us at that time. In my view, however, we need not and should not decide the question today.
*856II
Under the Eighth Amendment, the death penalty has been treated differently from all other punishments. See, e. g., California v. Ramos, 463 U. S. 992, 998-999, and n. 9 (1983). Among the most important and consistent themes in this Court’s death penalty jurisprudence is the need for special care and deliberation in decisions that may lead to the imposition of that sanction. The Court has accordingly imposed a series of unique substantive and procedural restrictions designed to ensure that capital punishment is not imposed without the serious and calm reflection that ought to precede any decision of such gravity and finality.
The restrictions that we have required under the Eighth Amendment affect both legislatures and the sentencing authorities responsible for decisions in individual cases. Neither automatic death sentences for certain crimes, for example, nor statutes committing the sentencing decision to the unguided discretion of judges or juries, have been upheld. See, e. g., Woodson v. North Carolina, 428 U. S. 280 (1976); Roberts v. Louisiana, 428 U. S. 325 (1976); Gregg v. Georgia, 428 U. S. 153, 188-189 (1976) (opinion of Stewart, Powell, and Stevens, JJ.) (discussing Furman v. Georgia, supra). We have rejected both legislative restrictions on the mitigating evidence that a sentencing authority may consider, e. g., Lockett v. Ohio, 438 U. S. 586 (1978); Eddings v. Oklahoma, 455 U. S. 104 (1982), and the lack of sufficiently precise restrictions on the aggravating circumstances that may be considered, e. g., Godfrey v. Georgia, 446 U. S. 420 (1980). As a practical matter we have virtually required that the death penalty be imposed only when a guilty verdict has been followed by separate trial-like sentencing proceedings, and we have extended many of the procedural restrictions applicable during criminal trials into these proceedings. See, e. g., Gardner v. Florida, 430 U. S. 349 (1977); Estelle v. Smith, 451 U. S. 454 (1981); Bullington v. Missouri, 451 U. S. 430 *857(1981). Legislatures have been forbidden to authorize capital punishment for certain crimes. Coker v. Georgia, 433 U. S. 584 (1977); Enmund v. Florida, 458 U. S. 782 (1982); see also Ford v. Wainwright, 477 U. S. 399 (1986) (Eighth Amendment forbids the execution of insane prisoners). Constitutional scrutiny in this area has been more searching than in the review of noncapital sentences. See Enmund v. Florida, supra, at 815, n. 27 (O’Connor, J., dissenting); Rummel v. Estelle, 445 U. S. 263, 272 (1980).
The case before us today raises some of the same concerns that have led us to erect barriers to the imposition of capital punishment in other contexts. Oklahoma has enacted a statute that authorizes capital punishment for murder, without setting any minimum age at which the commission of murder may lead to the imposition of that penalty. The State has also, but quite separately, provided that 15-year-old murder defendants may be treated as adults in some circumstances. Because it proceeded in this manner, there is a considerable risk that the Oklahoma Legislature either did not realize that its actions would have the effect of rendering 15-year-old defendants death eligible or did not give the question the serious consideration that would have been reflected in the explicit choice of some minimum age for death eligibility. Were it clear that no national consensus forbids the imposition of capital punishment for crimes committed before the age of 16, the implicit nature of the Oklahoma Legislature’s decision would not be constitutionally problematic. In the peculiar circumstances we face today, however, the Oklahoma statutes have presented this Court with a result that is of very dubious constitutionality, and they have done so without the earmarks of careful consideration that we have required for other kinds of decisions leading to the death penalty. In this unique situation, I am prepared to conclude that petitioner and others who were below the age of 1& at the time of their offense may not be executed under the authority of a capital punishment statute that specifies no mini*858mum age at which the commission of a capital crime can lead to the offender’s execution.*
The conclusion I have reached in this unusual case is itself unusual. 1 believe, however, that it is in keeping with the principles that have guided us in other Eighth Amendment cases. It is also supported by the familiar principle — applied in different ways in different contexts — according to which we should avoid unnecessary, or unnecessarily broad, constitutional adjudication. See generally, e. g., Ashwander v. TVA, 297 U. S. 288, 341-356 (1936) (Brandéis, J., concurring). The narrow conclusion I have reached in this case is consistent with the underlying rationale for -that principle, which was articulated many years ago by Justice Jackson: “We are not final because we are infallible, but we are infallible only because we are final.” Brown v. Allen, 344 U. S. 443, 540 (1953) (opinion concurring in result); see also Califano v. Yamasaki, 442 U. S. 682, 692-693 (1979). By leaving open for now the broader Eighth Amendment question that both the plurality and the dissent would resolve, the approach I take allows the ultimate moral issue at stake in the constitutional question to be addressed in the first in*859stance.by those best suited to do so, the people’s elected representatives.
■ For the reasons stated in' this opinion, I agree that petitioner’s death sentence should be vacated, and I therefore concur in the judgment of the Court.
Contrary to the dissent’s suggestion, the conclusion I have reached in this case does not imply that I would reach a similar conclusion in cases involving “those of extremely low’ intelligence, or those over 75, or any number of other appealing groups as to w'hich the existence of a national consensus regarding capital punishment may be in doubt. . . because they are not specifically named in the capital statutes.’’ See post, at 877. In this case, there is significant affirmative evidence of a national consensus forbidding the execution of defendants who were below the age of 16 at the time of the offense. The evidence includes 18 state statutes setting a minimum age of 16 or more, and it is such evidence — not the mere failure of Oklahoma to specify a minimum age or the “appealing” nature of the group to which petitioner belongs — that leaves me unwilling to conclude that petitioner may constitutionally be executed. Cases in which similarly persuasive evidence was lacking would in my view' not be analogous to the case before us today. The dissent is mistaken both when it reads into my discussion a contrary implication and when it suggests that there are ulterior reasons behind the implication it has incorrectly drawn.