dissenting.
The proper framework for analyzing respondent’s claims is not in doubt. First, we must determine whether the Massachusetts statute criminalizes expression protected by the First Amendment. If it does, then we must decide whether Massachusetts has a compelling interest in regulating that expression. To the extent that the Commonwealth’s interest does not justify the suppression of all protected conduct prohibited by the statute, we must further ask whether the law’s overbreadth is “not only . . . real, but substantial as well, judged in relation to the statute’s plainly legitimate sweep,” Broadrick v. Oklahoma, 413 U. S. 601, 615 (1973), once we have adopted any available narrowing constructions or severed offensive portions insofar as it lies within our power to do so. If the statute is excessively overbroad, we have no choice but to strike it down on its face, notwithstanding its laudable objectives and its numerous permissible applications; if it is not, then Oakes and others charged under *591it may argue only that their actions, though forbidden by the statute, may not constitutionally be proscribed.1
With the possible exception of the final step in this analysis, the resolution of these questions is straightforward. Photography, painting, and other two-dimensional forms of artistic reproduction described in Mass. Gen. Laws § 272:29A (1986) are plainly expressive activities that ordinarily qualify for First Amendment protection. See, e. g., Miller v. California, 413 U. S. 15 (1973) (works which, taken as a whole, possess serious artistic value are protected). And modeling, both independently and by virtue of its close association with those activities, enjoys like shelter under the First Amendment. , Cf. Schad v. Mount Ephraim, 452 U. S. 61, 66 (1981) (“[N]ude dancing is not without its First Amendment protections from official regulation”). Visual depictions of children engaged in live sexual performances or lewdly exhibiting their genitals cannot, of course, claim protected status, even though those depictions are not obscene. See New York v. Ferber, 458 U. S. 747 (1982). But other nonobscene representations of minors, including some that are pornographic, are shielded by the Constitution’s guarantee of free speech. Id., at 764-765. In particular, “nudity, without more is protected expression.” Id., at 765, n. 18, citing Erznoznik v. City of Jacksonville, 422 U. S. 205, 213 (1975). Because *592§ 29A’s prohibition extends to posing or exhibiting children “in a state of nudity,” rather than merely to their participation in live or simulated sexual conduct, the statute clearly restrains expression within the ambit of the First Amendment.
It is equally evident that the Commonwealth’s asserted interest in preventing the sexual exploitation and abuse of minors is “of surpassing importance.” Ferber, supra, at 757. See also Ginsberg v. New York, 390 U. S. 629, 639-641 (1968). The coercive enlistment, both overt and subtle, of children in the production of pornography is a grave and widespread evil which the States are amply justified in seeking to eradicate. Massachusetts’ interest in ending such conduct undoubtedly suffices to sustain the statute’s ban on encouraging, causing, or permitting persons one has reason to know are under 18 years of age to engage in any live sexual performance or any act that represents sexual conduct, for the purpose of visual representation or reproduction.
The Commonwealth lacks an overriding interest, however, in prohibiting adults from allowing minors to appear naked in photographs, films, and pictures with their genitals or, in the case of adolescent girls, their breasts less than opaquely covered under all circumstances except the production of such works “for a bona fide scientific or medical purpose, or for an educational or cultural purpose for a bona fide school, museum or library.” §29A. One situation where the Commonwealth’s interest falls glaringly short was cited by the Massachusetts Supreme Judicial Court: parents might want to photograph their infant children or toddlers in the bath or romping naked on the beach, yet § 29A threatens them with a prison term of between 10 and 20 years or a minimum fine of $10,000 for doing so. And §29A imposes those penalties even though parents have the same First Amendment interest in taking those photographs as they do in keeping a diary or boasting of their children’s antics, and even though their children would not thereby be harmed. Amicus American Sunbathing Association, a nudist organization with 30,000 *593members in the United States and Canada, further notes that family photographs taken by its members would subject them to possible prosecution, notwithstanding the protected character of their activity and their denial of any intrinsic connection between public nudity and shame. Massachusetts likewise lacks a compelling interest in forbidding nonexploitative films or photographs of topless adolescents — for instance, the poolside shots that are the norm rather than the exception along the Mediterranean seaboard, and that occur with some frequency on this side of the Atlantic as well — or in barring acting or professional modeling by teenagers that does not involve sexually explicit conduct.
In my view, the First Amendment also blocks the prohibition of nude posing by minors in connection with the production of works of art not depicting lewd behavior and not specifically prepared, in accordance with § 29A’s exclusion, for museums or libraries. Many of the world’s great artists — Degas, Renoir, Donatello,2 to name but a few — have worked from models under 18 years of age, and many acclaimed photographs and films have included nude or partially clad minors.3 The First Amendment rights of models, actors, artists, photographers, and filmmakers are surely not overborne by the Commonwealth’s interest in protecting minors from the risk of sexual abuse and exploitation, especially in view of the comprehensive set of laws targeted at those evils.4
*594Given that §29A is demonstrably overbroad, the next question is whether it fairly admits of a narrowing construction or whether offending portions of the statute might be severed, leaving its legitimate core prohibition intact. The answer to this question is that a restrictive reading of the statute or its partial invalidation is beyond our power. When we sit to review a decision resting on a state court’s construction of a state statute, that construction is binding on us, regardless of whether in its absence we would have read the statute in the same way or would have pruned it back before passing judgment. Ferber, 458 U. S., at 769, n. 24; Erznoznik, 422 U. S., at 216. “[W]e will not rewrite a state law to conform it to constitutional requirements.” Virginia v. American Booksellers Assn., Inc., 484 U. S. 383, *595397 (1988) (certifying interpretive questions to Virginia Supreme Court before ruling on First Amendment facial attack). In this case, § 29A’s prohibition on causing or allowing a minor to pose naked is unambiguous, and the Massachusetts Supreme Judicial Court expressly held that it forbids the various forms of constitutionally protected conduct just described. 401 Mass. 602, 605, 518 N. E. 2d 836, 838 (1988). In addition, although the phrase “to pose or be exhibited in a state of nudity” might easily have been excised, the court refused to sever and delete it, over the protest of three dissenters. Id., at 611, n. 4, 518 N. E. 2d, at 841, n. 4 (O’Connor, J., dissenting). We have no choice but to accept these authoritative pronouncements in adjudging the validity of §29A.
The test we employ is familiar. Because “conduct and not merely speech is involved, . . . the overbreadth of a statute must not only be real, but substantial as well, judged in relation to the statute’s plainly legitimate sweep.” Broadrick, 413 U. S., at 615. See also, e. g., Board of Airport Comm’rs of Los Angeles v. Jews for Jesus, Inc., 482 U. S. 569, 574 (1987); Houston v. Hill, 482 U. S. 451, 458 (1987); Ferber, supra, at 769. We will not topple a statute merely because we can conceive of a few impermissible applications. See City Council of Los Angeles v. Taxpayers for Vincent, 466 U. S. 789, 800, and n. 19 (1984). The possibility of a substantial number of realistic applications in contravention of the First Amendment, however, suffices to overturn a statute on its face. In this regard, it bears emphasizing that “the penalty to be imposed is relevant in determining whether demonstrable overbreadth is substantial.” Ferber, 458 U. S., at 773. Although “the fact that a criminal prohibition is involved does not obviate the need for the inquiry or a priori warrant a finding of substantial overbreadth,” ibid., it does appreciably shrink the amount of-overbreadth we will find constitutionally tolerable, particularly when the penalty is se*596vere. See also Houston v. Hill, supra, at 459 (“Criminal statutes must be scrutinized with particular care”).5
In this case, there is no gainsaying the gravity of the penalties meted out for violations of §29A. Infractions carry a fine of between $10,000 and $50,000, a prison term of between 10 and 20 years, or both. Respondent was himself sentenced to 10 years’ imprisonment for taking fewer than a dozen snapshots of his stepdaughter, which he apparently showed no one except the complainant. The severity of these sanctions significantly reduces the degree of overbreadth that the Constitution permits.
One can also readily adduce actual examples of protected conduct within §29A’s compass. Parents photograph their children without abusing them sexually in Massachusetts as elsewhere. The arts flourish there. Four nudist clubs affiliated with the American Sunbathing Association alone have been established in the Commonwealth, Brief for American Sunbathing Association as Amicus Curiae 2, and there may well be others.
The only question that might give one pause is whether the statute’s overbreadth is substantial. Unhappily, our precedents provide limited guidance in resolving this issue, because substantiality cannot be defined with exactitude and *597little overlap exists between the factual situations presented in our previous overbreadth cases and the circumstances confronting us here. But several considerations that have led us to strike down laws by reason of overbreadth tug with equal force in this case, strongly suggesting that § 29A cannot stand as it was written at the time respondent photographed his stepdaughter.
In Houston v. Hill, supra, at 464-466, we asked whether the sweeping nature of an ordinance making it a criminal offense to oppose, abuse, or interrupt a policeman in the performance of his duties was essential to achieve its ends, or whether a more narrowly tailored law could have attained the same objectives without abridging First Amendment freedoms to the same extent. Our finding that the law could have been drafted more tightly without sacrificing the achievement of its legitimate purposes impelled us to pronounce it fatally overbroad. Section 29A suffers from the same flaw. Its blanket prohibition on permitting minors to pose nude or employing nude models, without regard to the adult’s intentions or the sexually explicit nature of the minor’s conduct, nets a considerable amount of protected conduct. The statute can, moreover, easily be truncated. As the plurality describes, ante, at 582-583, and n. 2, Massachusetts itself has recently amended § 29A to lessen its threat to protected conduct by requiring that an adult act with “lascivious intent” to come within the statute’s prohibition.6 Mass. Gen. Laws §272:29A(a) (Supp. 1988). Alternatively, the Commonwealth could have followed the advice offered by the Justice Department in 1977. In considering legislation designed to combat the sexual exploitation of children in photographs and films, the House of Representatives initially considered banning the interstate dissemination, or the taking of photographs with intent or reason to know that they will be transported in interstate commerce, of children in a state of *598“ ‘nudity. . . depicted for the purpose of sexual stimulation or gratification of any individual who may view such depiction.’ ” See H. R. Rep. No. 95-696, p. 21 (1977) (quoting H. R. 4571, 95th Cong., 1st Sess. (1977)). The Justice Department opposed the inclusion of this provision on the ground that “it would be difficult to determine by what standard the ‘sexual stimulation or gratification’ could be assessed.” H. R. Rep. No. 95-696, at 21 (statement of Deputy Assistant Attorney General Keeney). The Justice Department suggested that “lewd exhibition of the genitals” be used in its place, ibid., and the House heeded that recommendation. Massachusetts could have followed the same course and modified § 29A’s reference to simple nudity, thereby aligning the law with the New York statute we upheld in Ferber. The availability of such simple correctives renders the statute’s overbreadth less acceptable.
Together with the stern sanctions § 29A imposes, the ease with which its unconstitutional applications might be eliminated lowers the hurdle respondent must clear in proving substantial overbreadth. By the standards set in our earlier decisions, that proof has in my judgment been made. The abundance of baby and child photographs taken every day without full frontal covering, not to mention the work of artists and filmmakers and nudist family snapshots, allows one to say, as the Court said in Houston v. Hill, 482 U. S., at 466-467 (citation omitted), that “[t]he ordinance’s plain language is admittedly violated scores of times daily, yet only some individuals — those chosen by the police in their unguided discretion — are arrested.”
Indeed, even if I were less confident that the statute was routinely violated by protected conduct — and the test, of course, is the relative frequency of such violations, not what we believe is the likelihood that such violations will in fact be prosecuted — I would reach the same conclusion. In Erznoznik v. City of Jacksonville, 422 U. S. 205 (1975), we struck down for overbreadth a statute making it a public nuisance to *599show films at a drive-in theater displaying bare buttocks, pubic areas, or female breasts, if the screen was visible from a public area. By way of justification we said: “[The statute] would bar a film containing a picture of a baby’s buttocks, the nude body of a war victim, or scenes from a culture in which nudity is indigenous. The ordinance also might prohibit newsreel scenes of the opening of an art exhibit as well as shots of bathers on a beach. Clearly all nudity cannot be deemed obscene even as to minors.” Id., at 213. We saw no reason to inquire into the frequency with which such scenes appeared at drive-in movies in Jacksonville; the fact that they might be shown, and sometimes were shown, was enough. The amount of protected conduct that occurs and quite plainly is covered by § 29A is undoubtedly far greater than the speculative occurrences we found sufficient to establish substantial overbreadth in Erznoznik, where, in addition, the attendant penalties were puny by comparison. Thus, even granted a stingy estimate of the extent of § 29A’s over-breadth, the statute must fall. I would affirm the decision of the Massachusetts Supreme Judicial Court.
I agree with Justice Scalia that a State cannot salvage a criminal conviction under a law found to be overbroad, or safeguard its right to prosecute under a law challenged as overbroad, by curing the statute’s adjudicated or alleged infirmity prior to review of that conviction or ruling of statutory invalidation by the highest reviewing court. The deterrent effect of the overbreadth doctrine would be significantly impaired if this avenue were open to the States, for oftentimes the strongest and earliest attacks on overbroad laws are, not surprisingly, brought by criminal defendants. Accordingly, I join Part I of Justice Scalia’s opinion holding that a defendant’s overbreadth challenge cannot be rendered moot by narrowing the statute after the conduct for which he has been indicted occurred — the only proposition to which five Members of the Court have subscribed in this case.
See, e. g., R. Thomson, Degas, The Nudes 40-53, 119-125 (1988); K. Clark, The Nude 48-49, 154-161 (1956).
Numerous contemporary examples of nonpornographic photographs, films, and paintings that would invite prosecution under § 29A if produced in Massachusetts but which almost certainly caused no harm to those depicted in them are collected in App. to Brief for Law & Humanities Institute as Amicus Curiae.
The utility of §29A in preventing sexual abuse and exploitation appears dubious when assessed against the backdrop of other statutes designed to achieve the same end. Massachusetts already has laws prohibiting assault and battery, Mass. Gen. Laws §265:13A (1986); indecent *594assault and battery of a child under 14, § 265:13B; indecent assault and battery of a child 14 or older, § 265:13H; rape, § 265:22; forcible rape of a child under 16, § 265:22A; rape and abuse of a child under 16, § 265:23; assault with intent to rape, § 265:24; drugging persons to commit unlawful sexual intercourse, §272:3; open and gross lewdness, §272:16; dissemination of matter harmful to minors, § 272:28; dissemination or possession with intent to disseminate visual material of child in state of nudity or engaged in sexual conduct, §272:29B; and unnatural and lascivious acts with a child under 16, § 272:35A. Virtually every prosecution under § 29A has also involved charges under several of these other statutes. See App. to Brief for Petitioner. The marginal deterrent effect of § 29A may therefore be slight, thereby reducing the Commonwealth’s interest as against the First Amendment interests in conflict with § 29A. Of course, the penalties for violating § 29A are high; in fact, however odd the underlying scale of values or predictions of deterrence may appear, the punishment for allowing a child to be photographed nude exceeds that for dissemination of matter harmful to minors, § 272:28, and unnatural and lascivious acts with a child under 16, § 272:35A, and includes a maximum prison term in excess of that for indecent assault and battery of a child under 14, §265:13B (20 years under §29A versus 10 years under §265:13B). It is questionable, however, what marginal difference the unusual stiffness of these penalties makes in forestalling the production of pornography or the sexual abuse of children, which are often punishable under separate statutes. Section 29A’s most significant deterrent effect may well be on constitutionally protected conduct.
In considering a facial challenge of this kind, we have no reason to decide, of coursé, whether respondent’s own conduct may legitimately be proscribed. Nor is it for us to say what exactly Oakes did when the evidence is sharply conflicting, particularly when we are remanding the case for further consideration of his as-applied challenge. Justice Scalia’s statement that “the defendant here apparently intended to send his stepdaughter’s photograph” to one of the “pornographic magazines that use young female models,” ante, at 588, therefore seems to me inappropriate. The only record support for this assertion of which I am aware is the complainant’s testimony at trial, ambiguous with respect to Oakes’ intentions regarding the photographs at issue here, that “[h]e wanted to make me big for Playboy Magazine.” App. 25a. In any event, nothing this Court says should be taken to constrain the power of the Massachusetts courts to determine facts on remand.
I venture no views as to the constitutionality of § 29A as amended.