dissenting.
Today the Court carries forward the “difficult and continuing venture in substantive due process,” Planned Parenthood of Central Missouri v. Danforth, 428 U. S. 52 (1976) *786(White, J., dissenting), that began with the decision in Roe v. Wade, 410 U. S. 113 (1973), and has led the Court further and further afield in the 13 years since that decision was handed down. I was in dissent in Roe v. Wade and am in dissent today. In Part I below, I state why I continue to believe that this venture has been fundamentally misguided since its inception. In Part II, I submit that even accepting Roe v. Wade, the concerns underlying that decision by no means command or justify the results reached today. Indeed, in my view, our precedents in this area, applied in a manner consistent with sound principles of constitutional adjudication, require reversal of the Court of Appeals on the ground that the provisions before us are facially constitutional.1
I
The rule of stare decisis is essential if case-by-case judicial decisionmaking is to be reconciled with the principle of the *787rule of law, for when governing legal standards are open to revision in every case, deciding cases becomes a mere exercise of judicial will, with arbitrary and unpredictable results. But stare decisis is not the only constraint upon judicial decisionmaking. Cases — like this one — that involve our assumed power to set aside on grounds of unconstitutionality a state or federal statute representing the democratically expressed will of the people call other considerations into play. Because the Constitution itself is ordained and established by the people of the United States, constitutional adjudication by this Court does not, in theory at any rate, frustrate the authority of the people to govern themselves through institutions of their own devising and in accordance with principles of their own choosing. But decisions that find in the Constitution principles or values that cannot fairly be read into that document usurp the people’s authority, for such decisions represent choices that the people have never made and that they cannot disavow through corrective legislation. For this reason, it is essential that this Court maintain the power to restore authority to its proper possessors by correcting constitutional decisions that, on reconsideration, are found to be mistaken.
The Court has therefore adhered to the rule that stare decisis is not rigidly applied in cases involving constitutional issues, see Glidden Co. v. Zdanok, 370 U. S. 530, 543 (1962) (opinion of Harlan, J.), and has not hesitated to overrule decisions, or even whole lines of cases, where experience, scholarship, and reflection demonstrated that their fundamental premises were not to be found in the Constitution. Stare decisis did not stand in the way of the Justices who, in the late 1930’s, swept away constitutional doctrines that had placed unwarranted restrictions on the power of the State and Federal Governments to enact social and economic legislation, see United States v. Darby, 312 U. S. 100 (1941); West Coast Hotel Co. v. Parrish, 300 U. S. 379 (1937). Nor did stare decisis deter a different set of Justices, some 15 years *788later, from rejecting the theretofore prevailing view that the Fourteenth Amendment permitted the States to maintain the system of racial segregation. Brown v. Board of Education, 347 U. S. 483 (1954). In both instances, history has been far kinder to those who departed from precedent than to those who would have blindly followed the rule of stare decisis. And only last Term, the author of today’s majority opinion reminded us once again that “when it has become apparent that a prior decision has departed from a proper understanding” of the Constitution, that decision must be overruled. Garcia v. San Antonio Metropolitan Transit Authority, 469 U. S. 528, 557 (1985).
In my view, the time has come to recognize that Roe v. Wade, no less than the cases overruled by the Court in the decisions I have just cited, “departs from a proper understanding” of the Constitution and to overrule it. I do not claim that the arguments in support of this proposition are new ones or that they were not considered by the Court in Roe or in the cases that succeeded it. Cf. Akron v. Akron Center for Reproductive Health, Inc., 462 U. S. 416, 419-420 (1983). But if an argument that a constitutional decision is erroneous must be novel in order to justify overruling that precedent, the Court’s decisions in Lochner v. New York, 198 U. S. 45 (1905), and Plessy v. Ferguson, 163 U. S. 537 (1896), would remain the law, for the doctrines announced in those decisions were nowhere more eloquently or incisively criticized than in the dissenting opinions of Justices Holmes (in Lochner) and Harlan (in both cases). That the flaws in an opinion were evident at the time it was handed down is hardly a reason for adhering to it.
A
Roe v. Wade posits that a woman has a fundamental right to terminate her pregnancy, and that this right may be restricted only in the service of two compelling state interests: the interest in maternal health (which becomes compelling *789only at the stage in pregnancy at which an abortion becomes more hazardous than carrying the pregnancy to term) and the interest in protecting the life of the fetus (which becomes compelling only at the point of viability). A reader of the Constitution might be surprised to find that it encompassed these detailed rules, for the text obviously contains no references to abortion, nor, indeed, to pregnancy or reproduction generally; and, of course, it is highly doubtful that the authors of any of the provisions of the Constitution believed that they were giving protection to abortion. As its prior cases clearly show, however, this Court does not subscribe to the simplistic view that constitutional interpretation can possibly be limited to the “plain meaning” of the Constitution’s text or to the subjective intention of the Framers. The Constitution is not a deed setting forth the precise metes and bounds of its subject matter; rather, it is a document announcing fundamental principles in value-laden terms that leave ample scope for the exercise of normative judgment by those charged with interpreting and applying it. In particular, the Due Process Clause of the Fourteenth Amendment, which forbids the deprivation of “life, liberty, or property without due process of law,” has been read by the majority of the Court to be broad enough to provide substantive protection against state infringement of a broad range of individual interests. See Moore v. East Cleveland, 431 U. S. 494, 541-552 (1977) (WHITE, J., dissenting).
In most instances, the substantive protection afforded the liberty or property of an individual by the Fourteenth Amendment is extremely limited: State action impinging on individual interests need only be rational to survive scrutiny under the Due Process Clause, and the determination of rationality is to be made with a heavy dose of deference to the policy choices of the legislature. Only “fundamental” rights are entitled to the added protection provided by strict judicial scrutiny of legislation that impinges upon them. See id., at 499 (opinion of Powell, J.); id., at 537 (Stewart, J., joined by *790Rehnquist, J., dissenting); id., at 547-549 (White, J., dissenting). I can certainly agree with the proposition — which I deem indisputable — that a woman’s ability to choose an abortion is a species of “liberty” that is subject to the general protections of the Due Process Clause. I cannot agree, however, that this liberty is so “fundamental” that restrictions upon it call into play anything more than the most minimal judicial scrutiny.
Fundamental liberties and interests are most clearly present when the Constitution provides specific textual recognition of their existence and importance. Thus, the Court is on relatively firm ground when it deems certain of the liberties set forth in the Bill of Rights to be fundamental and therefore finds them incorporated in the Fourteenth Amendment’s guarantee that no State may deprive any person of liberty without due process of law. When the Court ventures further and defines as “fundamental” liberties that are nowhere mentioned in the Constitution (or that are present only in the so-called “penumbras” of specifically enumerated rights), it must, of necessity, act with more caution, lest it open itself to the accusation that, in the name of identifying constitutional principles to which the people have consented in framing their Constitution, the Court has done nothing more than impose its own controversial choices of value upon the people.
Attempts to articulate the constraints that must operate upon the Court when it employs the Due Process Clause to protect liberties not specifically enumerated in the text of the Constitution have produced varying definitions of “fundamental liberties.” One approach has been to limit the class of fundamental liberties to those interests that are “implicit in the concept of ordered liberty” such that “neither liberty nor justice would exist if [they] were sacrificed.” Palko v. Connecticut, 302 U. S. 319, 325, 326 (1937); see Moore v. East Cleveland, 431 U. S., at 537 (Stewart, J., joined by Rehnquist, J., dissenting). Another, broader approach is *791to define fundamental liberties as those that are “deeply rooted in this Nation’s history and tradition.” Id., at 503 (opinion of Powell, J.); see also Griswold v. Connecticut, 381 U. S. 479, 501 (1965) (Harlan, J., concurring). These distillations of the possible approaches to the identification of unenumerated fundamental rights are not and do not purport to be precise legal tests or “mechanical yardstick^],” Poe v. Ullman, 367 U. S. 497, 544 (1961) (Harlan, J., dissenting). Their utility lies in their effort to identify some source of constitutional value that reflects not the philosophical predilections of individual judges, but basic choices made by the people themselves in constituting their system of government— “the balance struck by this country,” id., at 542 (emphasis added) — and they seek to achieve this end through locating fundamental rights either in the traditions and consensus of our society as a whole or in the logical implications of a system that recognizes both individual liberty and democratic order. Whether either of these approaches can, as Justice Harlan hoped, prevent “judges from roaming at large in the constitutional field,” Griswold, supra, at 502, is debatable. What for me is not subject to debate, however, is that either of the basic definitions of fundamental liberties, taken seriously, indicates the illegitimacy of the Court’s decision in Roe v. Wade.
The Court has justified the recognition of a woman’s fundamental right to terminate her pregnancy by invoking decisions upholding claims of personal autonomy in connection with the conduct of family life, the rearing of children, marital privacy, the use of contraceptives, and the preservation of the individual’s capacity to procreate. See Carey v. Population Services International, 431 U. S. 678 (1977); Moore v. East Cleveland, supra; Eisenstadt v. Baird, 405 U. S. 438 (1972); Griswold v. Connecticut, supra; Skinner v. Oklahoma, 316 U. S. 535 (1942); Pierce v. Society of Sisters, 268 U. S. 510 (1925); Meyer v. Nebraska, 262 U. S. 390 (1923). Even if each of these cases was correctly decided *792and could be properly grounded in rights that are “implicit in the concept of ordered liberty” or “deeply rooted in this Nation’s history and tradition,” the issues in the cases cited differ from those at stake where abortion is concerned. As the Court appropriately recognized in Roe v. Wade, “[t]he pregnant woman cannot be isolated in her privacy,” 410 U. S., at 159; the termination of a pregnancy typically involves the destruction of another entity: the fetus. However one answers the metaphysical or theological question whether the fetus is a “human being” or the legal question whether it is a “person” as that term is used in the Constitution, one must at least recognize, first, that the fetus is an entity that bears in its cells all the genetic information that characterizes a member of the species homo sapiens and distinguishes an individual member of that species from all others, and second, that there is no nonarbitrary line separating a fetus from a child or, indeed, an adult human being. Given that the continued existence and development — that is to say, the life — of such an entity are so directly at stake in the woman’s decision whether or not to terminate her pregnancy, that decision must be recognized as sui generis, different in kind from the others that the Court has protected under the rubric of personal or family privacy and autonomy.2 Accordingly, the *793decisions cited by the Court both in Roe and in its opinion today as precedent for the fundamental nature of the liberty to choose abortion do not, even if all are accepted as valid, dictate the Court’s classification.
If the woman’s liberty to choose an abortion is fundamental, then, it is not because any of our precedents (aside from Roe itself) command or justify that result; it can only be because protection for this unique choice is itself “implicit in the concept of ordered liberty” or, perhaps, “deeply rooted in this Nation’s history and tradition.” It seems clear to me that it is neither. The Court’s opinion in Roe itself convincingly refutes the notion that the abortion liberty is deeply rooted in the history or tradition of our people, as does the continuing and deep division of the people themselves over the question of abortion. As for the notion that choice in the matter of abortion is implicit in the concept of ordered liberty, it seems apparent to me that a free, egalitarian, and democratic society does not presuppose any particular rule or set of rules with respect to abortion. And again, the fact that many men and women of good will and high commitment to constitutional government place themselves on both sides of the abortion controversy strengthens my own conviction that the values animating the Constitution do not compel rec*794ognition of the abortion liberty as fundamental. In so denominating that liberty, the Court engages not in constitutional interpretation, but in the unrestrained imposition of its own, extraconstitutional value preferences.3
B
A second, equally basic error infects the Court’s decision in Roe v. Wade. The detailed set of rules governing state restrictions on abortion that the Court first articulated in Roe and has since refined and elaborated presupposes not only that the woman’s liberty to choose an abortion is fundamental, but also that the State’s countervailing interest in protecting fetal life (or, as the Court would have it, “potential human life,” 410 U. S., at 159) becomes “compelling” only at the point at which the fetus is viable. As Justice O’Con-nor pointed out three years ago in her dissent in Akron v. Akron Center for Reproductive Health, Inc., 462 U. S., at 461, the Court’s choice of viability as the point at which the State’s interest becomes compelling is entirely arbitrary. The Court’s “explanation” for the line it has drawn is that the State’s interest becomes compelling at viability “because the fetus then presumably has the capacity of meaningful life outside the mother’s womb.” 410 U. S., at 163. As one critic *795of Roe has observed, this argument “mistakes a definition for a syllogism.” Ely, The Wages of Crying Wolf: A Comment on Roe v. Wade, 82 Yale L. J. 920, 924 (1973).
The governmental interest at issue is in protecting those who will be citizens if their lives are not ended in the womb. The substantiality of this interest is in no way dependent on the probability that the fetus may be capable of surviving outside the womb at any given point in its development, as the possibility of fetal survival is contingent on the state of medical practice and technology, factors that are in essence morally and constitutionally irrelevant. The State’s interest is in the fetus as an entity in itself, and the character of this entity does not change at the point of viability under conventional medical wisdom. Accordingly, the State’s interest, if compelling after viability, is equally compelling before viability.4
*796c
Both the characterization of the abortion liberty as fundamental and the denigration of the State’s interest in preserving the lives of nonviable fetuses are essential to the detailed set of constitutional rules devised by the Court to limit the States’ power to regulate abortion. If either or both of these facets of Roe v. Wade were rejected, a broad range of limitations on abortion (including outright prohibition) that are now unavailable to the States would again become constitutional possibilities.
In my view, such a state of affairs would be highly desirable from the standpoint of the Constitution. Abortion is a hotly contested moral and political issue. Such issues, in our society, are to be resolved by the will of the people, either as expressed through legislation or through the general principles they have already incorporated into the Constitution they have adopted.6 Roe v. Wade implies that the people *797have already resolved the debate by weaving into the Constitution the values and principles that answer the issue. As I have argued, I believe it is clear that the people have never — not in 1787, 1791, 1868, or at any time since — done any such thing. I would return the issue to the people by overruling Roe v. Wade.
II
As it has evolved in the decisions of this Court, the freedom recognized by the Court in Roe v. Wade and its progeny is essentially a negative one, based not on the notion that abortion is a good in itself, but only on the view that the legitimate goals that may be served by state coercion of private choices regarding abortion are, at least under some circumstances, outweighed by the damage to individual autonomy and privacy that such coercion entails. In other words, the evil of abortion does not justify the evil of forbidding it. Cf. Stanley v. Georgia, 394 U. S. 557 (1969). But precisely because Roe v. Wade is not premised on the notion that abortion is itself desirable (either as a matter of constitutional entitlement or of social policy), the decision does not command the States to fund or encourage abortion, or even to approve *798of it. Rather, we have recognized that the States may legitimately adopt a policy of encouraging normal childbirth rather than abortion so long as the measures through which that policy is implemented do not amount to direct compulsion of the woman’s choice regarding abortion. Harris v. McRae, 448 U. S. 297 (1980); Maher v. Roe, 432 U. S. 464 (1977); Beal v. Doe, 432 U. S. 438 (1977). The provisions before the Court today quite obviously represent the State’s effort to implement such a policy.
The majority’s opinion evinces no deference toward the State’s legitimate policy. Rather, the majority makes it clear from the outset that it simply disapproves of any attempt by Pennsylvania to legislate in this area. The history of the state legislature’s decade-long effort to pass a constitutional abortion statute is recounted as if it were evidence of some sinister conspiracy. See ante, at 751-752. In fact, of course, the legislature’s past failure to predict the evolution of the right first recognized in Roe v. Wade is understandable and is in itself no ground for condemnation. Moreover, the legislature’s willingness to pursue permissible policies through means that go to the limits allowed by existing precedents is no sign of mens rea. The majority, however, seems to find it necessary to respond by changing the rules to invalidate what before would have seemed permissible. The result is a decision that finds no justification in the Court’s previous holdings, departs from sound principles of constitutional and statutory interpretation, and unduly limits the State’s power to implement the legitimate (and in some circumstances compelling) policy of encouraging normal childbirth in preference to abortion.
A
The Court begins by striking down statutory provisions designed to ensure that the woman’s choice of an abortion is fully informed — that is, that she is aware not only of the reasons for having an abortion, but also of the risks associated with an abortion and the availability of assistance that might *799make the alternative of normal childbirth more attractive than it might otherwise appear. At first blush, the Court’s action seems extraordinary: after all, Roe v. Wade purports to be about freedom of choice, and statutory provisions requiring that a woman seeking an abortion be afforded information regarding her decision not only do not limit her ability to choose abortion, but' also would appear to enhance her freedom of choice by helping to ensure that her decision whether or not to terminate her pregnancy is an informed one. Indeed, maximization of the patient’s freedom of choice — not restriction of his or her liberty — is generally perceived to be the principal value justifying the imposition of disclosure requirements upon physicians:
“The root premise is the concept, fundamental in American jurisprudence, that ‘[ejvery human being of adult years and sound mind has a right to determine what shall be done with his own body. . . . ’ True consent to what happens to one’s self is the informed exercise of a choice, and that entails an opportunity to evaluate knowledgeably the options available and the risks attendant upon each. The average patient has little or no understanding of the medical arts, and ordinarily has only his physician to whom he can look for enlightenment with which to reach an intelligent decision. From these almost axiomatic considerations springs the need, and in turn the requirement, of a reasonable divulgence by physician to patient to make such a decision possible.” Canterbury v. Spence, 150 U. S. App. D. C. 263, 271, 464 F. 2d 772, 780 (1972).
One searches the majority’s opinion in vain for a convincing reason why the apparently laudable policy of promoting informed consent becomes unconstitutional when the subject is abortion. The majority purports to find support in Akron v. Akron Center for Reproductive Health, Inc., 462 U. S. 416 (1983). But Akron is not controlling. The informed-*800consent provisions struck down in that case, as characterized by the majority, required the physician to advance tendentious statements concerning the unanswerable question of when human life begins, to offer merely speculative descriptions of the anatomical features of the fetus carried by the woman seeking the abortion, and to recite a “parade of horribles” suggesting that abortion is “a particularly dangerous procedure.” Id., at 444-445. I have no quarrel with the general proposition, for which I read Akron to stand, that a campaign of state-promulgated disinformation cannot be justified in the name of “informed consent” or “freedom of choice.” But the Pennsylvania statute before us cannot be accused of sharing the flaws of the ordinance at issue in Akron. As the majority concedes, the statute does not, on its face, require that the patient be given any information that is false or unverifiable. Moreover, it is unquestionable that all of the information required would be relevant in many cases to a woman’s decision whether or not to obtain an abortion.
Why, then, is the statute unconstitutional? The majority’s argument, while primarily rhetorical, appears to offer three answers. First, the information that must be provided will in some cases be irrelevant to the woman’s decision. This is true. Its pertinence to the question of the statute’s constitutionality, however, is beyond me. Legislators are ordinarily entitled to proceed on the basis of rational generalizations about the subject matter of legislation, and the existence of particular cases in which a feature of a statute performs no function (or is even counterproductive) ordinarily does not render the statute unconstitutional or even constitutionally suspect. Only where the statute is subject to heightened scrutiny by virtue of its impingement on some fundamental right or its employment of a suspect classification does the imprecision of the “fit” between the statute’s ends and means become potentially damning. Here, there is nothing to trigger such scrutiny, for the statute does not di*801rectly infringe the allegedly fundamental right at issue — the woman’s right to choose an abortion. Indeed, I fail to see how providing a woman with accurate information — whether relevant or irrelevant — could ever be deemed to impair any constitutionally protected interest (even if, as the majority hypothesizes, the information may upset her). Thus, the majority’s observation that the statute may require the provision of irrelevant information in some cases is itself an irrelevancy.
Second, the majority appears to reason that the informed-consent provisions are invalid because the information they require may increase the woman’s “anxiety” about the procedure and even “influence” her in her choice. Again, both observations are undoubtedly true; 'but they by no means cast the constitutionality of the provisions into question. It is in the very nature of informed-consent provisions that they may produce some anxiety in the patient and influence her in her choice. This is in fact their reason for existence, and — provided that the information required is accurate and non-misleading — it is an entirely salutary reason. If information may reasonably affect the patient’s choice, the patient should have that information; and, as one authority has observed, “the greater the likelihood that particular information will influence [the patient’s] decision, the more essential the information arguably becomes for securing her informed consent.” Appleton, Doctors, Patients and the Constitution, 63 Wash. U. L. Q. 183, 211 (1985). That the result of the provision of information may be that some women will forgo abortions by no means suggests that providing the information is unconstitutional, for the ostensible objective of Roe v. Wade is not maximizing the number of abortions, but maximizing choice. Moreover, our decisions in Maher, Beal, and Harris v. McRae all indicate that the State may encourage women to make their choice in favor of childbirth rather than abortion, and the provision of accurate information regarding abortion *802and its alternatives is a reasonable and fair means of achieving that objective.
Third, the majority concludes that the informed-consent provisions are invalid because they “intrud[e] upon the discretion of the pregnant woman’s physician,” ante, at 762, violate “the privacy of the informed-consent dialogue between the woman and her physician,” ibid., and “officially structure]” that dialogue, ante, at 763. The provisions thus constitute “state medicine” that “infringes upon [the physician’s] professional responsibilities.” Ibid. This is nonsensical. I can concede that the Constitution extends its protection to certain zones of personal autonomy and privacy, see Griswold v. Connecticut, 381 U. S., at 502 (White, J., concurring in judgment), and I can understand, if not share, the notion that that protection may extend to a woman’s decision regarding abortion. But I cannot concede the possibility that the Constitution provides more than minimal protection for the manner in which a physician practices his or her profession or for the “dialogues” in which he or she chooses to participate in the course of treating patients. I had thought it clear that regulation of the practice of medicine, like regulation of other professions and of economic affairs generally, was a matter peculiarly within the competence of legislatures, and that such regulation was subject to review only for rationality. See, e. g., Williamson v. Lee Optical of Oklahoma, Inc., 348 U. S. 483 (1955).
Were the Court serious about the need for strict scrutiny of regulations that infringe on the “judgment” of medical professionals, “structure” their relations with their patients, and amount to “state medicine,” there is no telling how many state and federal statutes (not to mention principles of state tort law) governing the practice of medicine might be condemned. And of course, there would be no reason why a concern for professional freedom could be confined to the medical profession: nothing in the Constitution indicates a preference for the liberty of doctors over that of lawyers, *803accountants, bakers, or brickmakers. Accordingly, if the State may not “structure” the dialogue between doctor and patient, it should also follow that the State may not, for example, require attorneys to disclose to their clients information concerning the risks of representing the client in a particular proceeding. Of course, we upheld such disclosure requirements only last Term. See Zauderer v. Office of Disciplinary Counsel, 471 U. S. 626 (1985).
The rationale for state efforts to regulate the practice of a profession or vocation is simple: the government is entitled not to trust members of a profession to police themselves, and accordingly the legislature may for the most part impose such restrictions on the practice of a profession or business as it may find necessary to the protection of the public. This is precisely the rationale for infringing the professional freedom of doctors by imposing disclosure requirements upon them: “Respect for the patient’s right of self-determination on particular therapy demands a standard set by law for physicians rather than one which physicians may or may not impose upon themselves.” Canterbury v. Spence, 150 U. S. App. D. C., at 275, 464 F. 2d, at 784. Unless one is willing to recast entirely the law with respect to the legitimacy of state regulation of professional conduct, the obvious rationality of the policy of promoting informed patient choice on the subject of abortion must defeat any claim that the disclosure requirements imposed by Pennsylvania are invalid because they infringe on “professional freedom” or on the “physician-patient relationship.”
I do not really believe that the Court’s invocation of professional freedom signals a retreat from the principle that the Constitution is largely unconcerned with the substantive aspects of governmental regulation of professional and business relations. Clearly, the majority is uninterested in undermining the edifice of post-New Deal constitutional law by extending its holding to cases that do not concern the issue of abortion. But if one assumes, as I do, that the majority *804is unwilling to commit itself to the implications of that part of its rhetoric which smacks of economic due process rights for physicians, it becomes obvious that the talk of “infringement of professional responsibility” is mere window dressing for a holding that must stand or fall on other grounds. And because the informed-consent provisions do not infringe the essential right at issue — the right of the woman to choose to have an abortion — the majority’s conclusion that the provisions are unconstitutional is without foundation.
B
The majority’s decision to strike down the reporting requirements of the statute is equally extraordinary. The requirements obviously serve legitimate purposes. The information contained in the reports is highly relevant to the State’s efforts to enforce § 3210(a) of the statute, which forbids abortion of viable fetuses except when necessary to the mother’s health. The information concerning complications plainly serves the legitimate goal of advancing the state of medical knowledge concerning maternal and fetal health. See Planned Parenthood of Central Missouri v. Danforth, 428 U. S., at 80. Given that the subject of abortion is a matter of considerable public interest and debate (constrained to some extent, of course, by the pre-emptive effect of this Court’s ill-conceived constitutional decisions), the collection and dissemination of demographic information concerning abortions is clearly a legitimate goal of public policy. Moreover, there is little reason to believe that the required reports, though fairly detailed, would impose an undue burden on physicians and impede the ability of their patients to obtain abortions, as all of the information required would necessarily be readily available to a physician who had performed an abortion. Accordingly, under this Court’s prior decisions in this area, the reporting requirements are constitutional. Planned Parenthood Assn. of Kansas City, Missouri, Inc. v. Ashcroft, 462 U. S. 476, 486-490 (1983) (opinion of Pow*805ell, J.); id., at 505 (opinion of O’Connor, J.); Planned Parenthood of Central Missouri v. Danforth, supra, at 79-81.
Nonetheless, the majority strikes down the reporting requirements because it finds that notwithstanding the explicit statutory command that the reports be made public only in a manner ensuring anonymity, “the amount of information about [the patient] and the circumstances under which she had an abortion are so detailed that identification is likely,” ante, at 767, and that “identification is the obvious purpose of these extreme reporting requirements,” ibid. Where these “findings” come from is mysterious, to say the least. The Court of Appeals did not make any such findings on the record before it, and the District Court expressly found that “the requirements of confidentiality in § 3214(e) regarding the identity of both patient and physician prevent any invasion of privacy which could present a legally significant burden on the abortion decision.” 552 F. Supp. 791, 804 (ED Pa. 1982). Rather than pointing to anything in the record that demonstrates that the District Court’s conclusion is erroneous, the majority resorts to the handy, but mistaken, solution of substituting its own view of the facts and strikes down the statute.
I can accept the proposition that a statute whose purpose and effect are to allow harassment and intimidation of citizens for their constitutionally protected conduct is unconstitutional, but the majority’s action in striking down the Pennsylvania statute on this basis is procedurally and substantively indefensible. First, it reflects a complete disregard for the principle, embodied in Federal Rule of Civil Procedure 52(a), that an appellate court must defer to a trial court’s findings of facts unless those findings are clearly erroneous. The Rule is expressly applicable to findings of fact that constitute the grounds for a district court’s action granting or refusing a preliminary injunction, and, of course, the Rule limits this Court to the same degree as it does any other *806federal appellate court, see United States v. General Dynamics Corp., 415 U. S. 486 (1974).
Second, the majority has seriously erred in purporting to make a final determination of fact, conclusive of the constitutionality of the statute, on a motion for preliminary injunction. In so doing, the Court overlooks the principle that although a district court’s findings of fact on a motion for a preliminary injunction are entitled to deference on appeal from the grant or denial of preliminary relief, “the findings of fact . . . made by a court granting a preliminary injunction are not binding at trial on the merits” because “a preliminary injunction is customarily granted on the basis of procedures that are less formal and evidence that is less complete than in a trial on the merits.” University of Texas v. Camenisch, 451 U. S. 390, 395 (1981) (emphasis added). What Camenisch stated to be true customarily is also true in this case: the record on which the motion for preliminary injunction was decided in the trial court consisted solely of affidavits and a stipulation of undisputed facts, none of which provides a sufficient basis for a conclusive finding on the complex question of the motive and effect of the reporting requirements and the adequacy of the statute’s protection of the anonymity of doctors and patients. Issuing what amounts to a final declaratory judgment on the constitutionality of the statute under these circumstances is highly inappropriate.
Finally, in addition to being procedurally flawed, the majority’s holding is substantively suspect. The information contained in the reports identifies the patient on the basis of age, race, marital status, and “political subdivision” of residence; the remainder of the information included in the reports concerns the medical aspects of the abortion. It is implausible that a particular patient could be identified on the basis of the combination of the general identifying information and the specific medical information in these reports by anyone who did not already know (at a minimum) that the woman had been pregnant and obtained an abortion. *807Accordingly, the provisions pose little or no threat to the woman’s privacy.
In sum, there is no basis here even for a preliminary injunction against the reporting provisions of the statute, much less for a final determination that the provisions are unconstitutional.
C
The majority resorts to linguistic nit-picking in striking down the provision requiring physicians aborting viable fetuses to use the method of abortion most likely to result in fetal survival unless that method would pose a “significantly greater medical risk to the life or health of the pregnant woman” than would other available methods. The majority concludes that the statute’s use of the word “significantly” indicates that the statute represents an unlawful “trade-off” between the woman’s health and the chance of fetal survival. Not only is this conclusion based on a wholly unreasonable interpretation of the statute, but the statute would also be constitutional even if it meant what the majority says it means.
The majority adopts the Court of Appeals’ view that the statute’s use of the term “significantly” renders it “ 'not susceptible to a construction that does not require the mother to bear an increased medical risk in order to save her viable fetus.’” Ante, at 769 (quoting 737 F. 2d 283, 300 (CA3 1984)). The term “significant” in this context, however, is most naturally read as synonymous with the terms “meaningful,” “cognizable,” “appreciable,” or “nonnegligible.” That is, the statute requires only that the risk be a real and identifiable one. Surely, if the State’s interest in preserving the life of a viable fetus is, as Roe purported to recognize, a compelling one, the State is at the very least entitled to demand that that interest not be subordinated to a purported maternal health risk that is in fact wholly insubstantial. The statute, on its face, demands no more than this of a doctor performing an abortion of a viable fetus.
*808Even if the Pennsylvania statute is properly interpreted as requiring a pregnant woman seeking abortion of a viable fetus'to endure a method of abortion chosen to protect the health of the fetus despite the existence of an alternative that in some substantial degree is more protective of her own health, I am not convinced that the statute is unconstitutional. The Court seems to read its earlier opinion in Colautti v. Franklin, 439 U. S. 379 (1979), as incorporating a holding that tradeoffs between the health of the pregnant woman and the survival of her viable fetus are constitutionally impermissible under Roe v. Wade. Of course, Colautti held no such thing: the Court there stated only that it did not address the “serious ethical and constitutional difficulties” that such a tradeoff would present. 439 U. S., at 400.6 Nothing in Colautti or any of the Court’s previous abortion decisions compels the per se “tradeoff” rule the Court adopts today.
The Court’s ruling in this respect is not even consistent with its decision in Roe v. Wade. In Roe, the Court conceded that the State’s interest in preserving the life of a viable fetus is a compelling one, and the Court has never disavowed that concession. The Court now holds that this compelling interest cannot justify any regulation that imposes a quantifiable medical risk upon the pregnant woman who seeks to abort a viable fetus: if attempting to save the fetus imposes any additional risk of injury to the woman, she must be permitted to kill it. This holding hardly accords with the usual understanding of the term “compelling interest,” which we have used to describe those governmental interests that are so weighty as to justify substantial and ordinarily impermissible impositions on the individual — impositions that, I had thought, could include the infliction of *809some degree of risk of physical harm. The most obvious illustration of this principle may be found in the opinion of the elder Justice Harlan in Jacobson v. Massachusetts, 197 U. S. 11, 29 (1905): “The liberty secured by the Fourteenth Amendment. . . consists, in part, in the right of a person ‘to live and work where he will,’ Allgeyer v. Louisiana, 165 U. S. 578; and yet he may be compelled, by force if need be, against his will and without regard to his personal wishes or his pecuniary interests, ... to take his place in the ranks of the army of his country and risk the chance of being shot down in its defense.” The actual holding of Jacobson provides another illustration, more pertinent to this particular case: the Court there sustained a regulation requiring all adult citizens of Cambridge, Massachusetts, to be vaccinated against smallpox, notwithstanding that exposure to vaccination carried with it a statistical possibility of serious illness and even death. If, as I believe these examples demonstrate, a compelling state interest may justify the imposition of some physical danger upon an individual, and if, as the Court has held, the State has a compelling interest in the preservation of the life of a viable fetus, I find the majority’s unwillingness to tolerate the imposition of any nonnegligible risk of injury to a pregnant woman in order to protect the life of her viable fetus in the course of an abortion baffling.
The Court’s ruling today that any tradeoff between the woman’s health and fetal survival is impermissible is not only inconsistent with Roe’s recognition of a compelling state interest in viable fetal life; it directly contradicts one of the essential holdings of Roe — that is, that the State may forbid all postviability abortions except when necessary to protect the life or health of the pregnant woman. As is evident, this holding itself involves a tradeoff between maternal health and protection of the fetus, for it plainly permits the State to forbid a postviability abortion even when such an abortion may be statistically safer than carrying the pregnancy to *810term, provided that the abortion is not medically necessary.7 The tradeoff contained in the Pennsylvania statute, even as interpreted by the majority, is no different in kind: the State has simply required that when an abortion of some kind is medically necessary, it shall be conducted so as to spare the fetus (to the greatest degree possible) unless a method less protective of the fetus is itself to some degree medically necessary for the woman. That this choice may involve the imposition of some risk on the woman undergoing the abortion should be no more troublesome than that a prohibition on nonnecessary postviability abortions may involve the imposition of some risk on women who are thereby forced to continue their pregnancies to term; yet for some reason, the Court concludes that whereas the tradeoffs it devises are compelled by the Constitution, the essentially indistinguishable tradeoff the State has attempted is foreclosed. This cannot be the law.
The framework of rights and interests devised by the Court in Roe v. Wade indicates that just as a State may prohibit a postviability abortion unless it is necessary to protect the life or health of the woman, the State may require that postviability abortions be conducted using the method most protective of the fetus unless a less protective method is necessary to protect the life or health of the woman. Under this standard, the Pennsylvania statute — which does not require the woman to accept any significant health risks to protect the fetus — is plainly constitutional.
D
The Court strikes down the statute’s second-physician requirement because, in its view, the existence of a medical emergency requiring an immediate abortion to save the life of the pregnant woman would not be a defense to a prosecution *811under the statute. The Court does not question the proposition, established in the Ashcroft case, that a second-physician requirement accompanied by an exception for emergencies is a permissible means of vindicating the compelling state interest in protecting the lives of viable fetuses. Accordingly, the majority’s ruling on this issue does not on its face involve a substantial departure from the Court’s previous decisions.
What is disturbing about the Court’s opinion on this point is not the general principle on which it rests, but the manner in which that principle is applied. The Court brushes aside the fact that the section of the statute in which the second-physician requirement is imposed states that “[i]t shall be a complete defense to any charge brought against a physician for violating the requirements of this section that he had concluded, in good faith, in his best medical judgment, . . . that the abortion was necessary to preserve maternal life or health” (emphasis added). 18 Pa. Cons. Stat. § 3210(a) (1982). This language is obviously susceptible of the construction the State advances: namely, that it is a defense to a charge of violating the second-physician requirement that the physician performing the abortion believed that performing an abortion in the absence of a second physician was necessary to the life or health of the mother.
The Court’s rejection of this construction is based on its conclusion that the statutory language “does not relate to the second-physician requirement” and that “its words are not words of emergency.” Ante, at 771. This reasoning eludes me. The defense of medical necessity “relates” to any charge that a doctor has violated one of the requirements of the section in which it appears, and the second-physician requirement is imposed by that section. The defense thus quite evidently “relates” to the second-physician requirement. True, the “words” of the defense are not “words of emergency,” but words of necessity. Why this should make a difference is unclear: a defense of medical necessity is fully as protective of the interests of the pregnant woman as a defense of *812“emergency.” The Court falls back, ibid., on the notion that the legislature “knows how to provide a medical-emergency exception when it chooses to do so.” No doubt. But the legislature obviously also “knows how” to provide a medical-necessity exception, and it has done so. Why this exception is insufficient is unexplained and inexplicable.
The Court’s rejection of a perfectly plausible reading of the statute flies in the face of the principle — which until today I had thought applicable to abortion statutes as well as to other legislative enactments — that “[wjhere fairly possible, courts should construe a statute to avoid a danger of unconstitutionality.” Planned Parenthood Assn. of Kansas City, Missouri, Inc. v. Ashcroft, 462 U. S., at 493. The Court’s reading is obviously based on an entirely different principle: that in cases involving abortion, a permissible reading of a statute is to be avoided at all costs. Not sharing this viewpoint, I cannot accept the majority’s conclusion that the statute does not provide for the equivalent of a defense of emergency.8
E
Finally, the majority refuses to vacate the preliminary injunction entered against the enforcement of the parental notice and consent provisions of the statute. See ante, at 758, n. 9. The reason offered is that the propriety of the injunction depends upon the adequacy of the rules, recently promulgated by the Pennsylvania Supreme Court, setting forth *813procedures by which a minor desiring an abortion may speedily and confidentially obtain either judicial approval of her decision to obtain an abortion or a judicial determination that she herself is capable of an informed consent to the procedure. The Court concludes that review of the rules is best carried out in the first instance in the District Court.
The Court’s decision in Ashcroft, however, compels the conclusion that the Third Circuit erred in directing that the operation of the parental notice and consent provisions be enjoined pending promulgation of the required rules; accordingly, the injunction should be vacated irrespective of the adequacy of those rules. As the Court of Appeals apparently recognized, the Pennsylvania statute, on its face, is substantively identical to that upheld by the Court in Ashcroft; thus, the sole basis for the injunction ordered by the Court of Appeals was the absence of procedural rules implementing the statute. What the Court of Appeals failed to recognize was that this Court denied relief to the plaintiffs challenging the statute in Ashcroft despite the same purported defect: in that case, as in this, the State Supreme Court had not yet promulgated rules establishing the expedited procedures called for by the statute. Nonetheless, as Justice Powell’s opinion explained, the plaintiffs were not entitled to any relief against enforcement of the statutory scheme, as “[t]here is no reason to believe that [the State] will not expedite any appeal consistent with the mandate in our prior opinions.” 462 U. S., at 491, n. 16. Similarly, there was no reason here for the Court of Appeals to believe that Pennsylvania would not provide for the adequate, expedited procedures contemplated by the statute; thus, its entry of an injunction against enforcement of the statute was erroneous.
Ill
The decision today appears symptomatic of the Court’s own insecurity over its handiwork in Roe v. Wade and the cases following that decision. Aware that in Roe it essen*814tially created something out of nothing and that there are many in this country who hold that decision to be basically illegitimate, the Court responds defensively. Perceiving, in a statute implementing the State’s legitimate policy of preferring childbirth to abortion, a threat to or criticism of the decision in Roe v. Wade, the majority indiscriminately strikes down statutory provisions that in no way contravene the right recognized in Roe. I do not share the warped point of view of the majority, nor can I follow the tortuous path the majority treads in proceeding to strike down the statute before us. I dissent.
I shall, for the most part, leave to one side the Court’s somewhat extraordinary procedural rulings. I do not strongly disagree with the Court’s decision to read a finality requirement into 28 U. S. C. § 1254(2), although I would have thought it incumbent on the Court to explain why the Court of Appeals’ judgment as to the statutory provisions before us today, which represents a definitive ruling on their constitutionality, is not sufficiently “final” to satisfy the jurisdictional statute as interpreted by the Court.
As for the Court’s ruling that it is permissible for an appellate court to resolve an appeal from the grant or the denial of a preliminary injunction by issuing a final judgment as to the constitutionality of a statute, I do not disagree that this may, in rare cases, be an appropriate course of action where the constitutional issues are clear. I would stress that this is by no means the preferred course of action in the run of cases, and I assume that the majority’s opinion is not to the contrary. I do disagree quite strongly with the majority's application of this principle here, as I believe, contrary to the majority, that it is quite evident that the statute before us is constitutional on its face. I also believe, as will become evident, that at least one of the Court’s rulings is exceedingly inappropriate in view of the preliminary posture of this case even if the majority’s legal premises are accepted.
That the abortion decision, like the decisions protected in Griswold, Eisenstadt, and Carey, concerns childbearing (or, more generally, family life) in no sense necessitates a holding that the liberty to choose abortion is “fundamental.” That the decision involves the destruction of the fetus renders it different in kind from the decision not to conceive in the first place. This difference does not go merely to the weight of the state interest in regulating abortion; it affects as well the characterization of the liberty interest itself. For if the liberty to make certain decisions with respect to contraception without governmental constraint is “fundamental,” it is not only because those decisions are “serious” and “important” to the individual, see ante, at 776 (Stevens, J., concurring), but also because some value of privacy or individual autonomy that is somehow implicit in the scheme of ordered liberties established by the Constitution supports a judgment that such decisions are none of government’s business. The *793same cannot be said where, as here, the individual is not “isolated in her privacy.”
My point can be illustrated by drawing on a related area in which fundamental liberty interests have been found: childrearing. The Court’s decisions in Moore v. East Cleveland, Pierce v. Society of Sisters, and Meyer v. Nebraska can be read for the proposition that parents have a fundamental liberty to make decisions with respect to the upbringing of their children. But no one would suggest that this fundamental liberty extends to assaults committed upon children by their parents. It is not the case that parents have a fundamental liberty to engage in such activities and that the State may intrude to prevent them only because it has a compelling interest in the well-being of children; rather, such activities, by their very nature, should be viewed as outside the scope of the fundamental liberty interest.
Justice Stevens asserts, ante, at 778, that I am “quite wrong in suggesting that the Court is imposing value preferences on anyone else” when it denominates the liberty to choose abortion as “fundamental” (in contradistinction to such other, nonfundamental liberties as the liberty to use dangerous drugs or to operate a business without governmental interference) and thereby disempowers state electoral majorities from legislating in this area. I can only respond that I cannot conceive of a definition of the phrase “imposing value preferences” that does not encompass the Court’s action.
Justice Stevens also suggests that it is the legislative majority that has engaged in “the unrestrained imposition of its own, extraconstitutional value preferences” when a state legislature restricts the availability of abortion. Ibid. But a legislature, unlike a court, has the inherent power to do so unless its choices are constitutionally forbidden, which, in my view, is not the case here.
Contrary to Justice Stevens’ suggestion, ibid,., this is no more a “theological” position than is the Court’s own judgment that viability is the point at which the state interest becomes compelling. (Interestingly, Justice Stevens omits any real effort to defend this judgment.) The point is that the specific interest the Court has recognized as compelling after the point of viability — that is, the interest in protecting “potential human life” — is present as well before viability, and the point of viability seems to bear no discernible relationship to the strength of that interest. Thus, there is no basis for concluding that the essential character of the state interest becomes transformed at the point of viability.
Further, it is self-evident that neither the legislative decision to assert a state interest in fetal life before viability nor the judicial decision to recognize that interest as compelling constitutes an impermissible “religious” decision merely because it coincides with the belief of one or more religions. Certainly the fact that the prohibition of murder coincides with one of the Ten Commandments does not render a State’s interest in its murder statutes less than compelling, nor are legislative and judicial decisions concerning the use of the death penalty tainted by their correspondence to varying religious views on that subject. The simple, and perhaps unfortunate, fact of the matter is that in determining whether to assert an interest in fetal life, a State cannot avoid taking a position that will correspond to some religious beliefs and contradict others. The same is true to some extent with respect to the choice this Court faces in characterizing an asserted state *796interest in fetal life, for denying that such an interest is a “compelling” one necessarily entails a negative resolution of the “religious” issue of the humanity of the fetus, whereas accepting the State’s interest as compelling reflects at least tolerance for a state decision that is congruent with the equally “religious” position that human life begins at conception. Faced with such a decision, the most appropriate course of action for the Court is to defer to a legislative resolution of the issue: in other words, if a state legislature asserts an interest in protecting fetal life, I can see no satisfactory basis for denying that it is compelling.
Justice Stevens, see ante, at 776-777, n. 4, finds a contradiction between my recognition that constitutional analysis requires more than mere textual analysis or a search for the specific intent of the Framers, supra, at 789, and my assertion that it is ultimately the will of the people that is the source of whatever values are incorporated in the Constitution. The fallacy of Justice Stevens’ argument is glaring. The rejection of what has been characterized as “clause-bound” interpretivism, J. Ely, Democracy and Distrust 12 (1980), does not necessarily carry with it a rejection of the notion that constitutional adjudication is a search for values and principles that are implicit (and explicit) in the structure of rights and institutions that the people have themselves created. The implications of those values for the resolution of particular issues will in many if not most cases not have been explicitly considered when the values themselves were *797chosen — indeed, there will be some cases in which those who framed the provisions incorporating certain principles into the Constitution will be found to have been incorrect in their assessment of the consequences of their decision. See, e. g., Brown v. Board of Education, 347 U. S. 483 (1954). Nonetheless, the hallmark of a correct decision of constitutional law is that it rests on principles selected by the people through their Constitution, and not merely on the personal philosophies, be they libertarian or authoritarian, of the judges of the majority. While constitutional adjudication involves judgments of value, it remains the case that some values are indeed “extraconstitutional,” in that they have no roots in the Constitution that the people have chosen. The Court’s decision in Lochner v. New York, 198 U. S. 45 (1905), was wrong because it rested on the Court’s belief that the liberty to engage in a trade or occupation without governmental regulation was somehow fundamental — an assessment of value that was unsupported by the Constitution. I believe that Roe v. Wade—and today’s decision as well — rests on similarly extraconstitutional assessments of the value of the liberty to choose an abortion.
Interestingly, the Court’s statement seems to have assumed that the Court would have had the same authority over “ethical questions” as “constitutional issues” had it chosen to reach them — an illuminating revelation of the state of the Court’s jurisprudence in this area.
Surely it cannot be argued that any abortion that is safer than delivery is medically necessary, since under such a definition an abortion would be medically necessary in all pregnancies.
Even if I were to accept the majority’s conclusion that the medical-necessity defense of § 3210(a) is not specifically applicable to charges brought under § 3210(c), I would not strike down the statute. Under Pennsylvania criminal law, justification is a defense, see 18 Pa. Cons. Stat. § 502 (1982), and, under the general rule of justification, conduct is deemed justified if “the actor believes [it] to be necessary to avoid a harm or evil to . . . another,” and “the harm or evil sought to be avoided by such conduct is greater than that sought to be prevented by the law defining the offense charged.” § 503(a)(1). I have little doubt that a Pennsylvania court applying this statute would find noncompliance with the second-physician rule justified where necessary to save the life of the pregnant woman.