concurring.
Although I join the Court’s opinion and judgment in this case, it seems to me that several factors merit more emphasis than it gives them. I therefore add the following remarks.
I
With respect to the remand to the Court of Appeals on the issue of official immunity from liability for monetary damages,1 it seems to me not entirely irrelevant that there was substantial evidence that Donaldson consistently refused treatment that was offered to him, claiming that he was not mentally ill and needed no treatment.2 *579The Court appropriately takes notice of the uncertainties of psychiatric diagnosis and therapy, and the reported cases are replete with evidence of the divergence of medical opinion in this vexing area. E. g., Greenwood v. United States, 350 U. S. 366, 375 (1956). See also Drope v. Missouri, 420 U. S. 162 (1975). Nonetheless, one of the few areas of agreement among behavioral specialists is that an uncooperative patient cannot benefit from therapy and that the first step in effective treatment is acknowledgment by the patient that he is suffering from an abnormal condition. See, e. g., Katz, The Right to Treatment — An Enchanting Legal Fiction? 36 U. Chi. L. Rev. 755, 768-769 (1969). Donaldson’s adamant refusal to do so should be taken into account in considering petitioner’s good-faith defense.
Perhaps more important to the issue of immunity is a factor referred to only obliquely in the Court’s opinion. On numerous occasions during the period of his confinement Donaldson unsuccessfully sought release in the Florida courts; indeed, the last of these proceedings was terminated only a few months prior to the bringing of this action. See 234 So. 2d 114 (1969), cert, denied, 400 U. S. 869 (1970). Whatever the reasons for the state courts’ repeated denials of relief, and regardless of whether they correctly resolved the issue tendered to them, petitioner and the other members of the medical staff at Florida State Hospital would surely have been justified in considering each such judicial decision as an approval of continued confinement and an independent intervening reason for continuing Donaldson’s custody. Thus, this fact is inescapably related to the issue of immunity and must be considered by the Court of Appeals on remand and, if a new trial on this issue is ordered, by the District Court.3
*580II
As the Court points out, ante, at 570 n. 6, the District Court instructed the jury in part that “a person who is involuntarily civilly committed to a mental hospital does have a constitutional right to receive such treatment as will give him a realistic opportunity to he cured” (emphasis added), and the Court of Appeals unequivocally approved this phrase, standing alone, as a correct statement of the law. 493 F. 2d 507, 520 (CA5 1974). The Court’s opinion plainly gives no approval to that holding and makes clear that it binds neither the parties to this case nor the courts of the Fifth Circuit. See ante, at 577-578, n. 12. Moreover, in light of its importance for future litigation in this area, it should be emphasized that the Court of Appeals’ analysis has no basis in the decisions of this Court.
A
There can be no doubt that involuntary commitment to a mental hospital, like involuntary confinement of an individual for any reason, is a deprivation of liberty which the State cannot accomplish without due process of law. Specht v. Patterson, 386 U. S. 605, 608 (1967). Cf. In re Gault, 387 U. S. 1, 12-13 (1967). Commitment must be justified on the basis of a legitimate state interest, and the reasons for committing a particular individual must be established in an appropriate proceeding. Equally important, confinement must cease when those reasons no longer exist. See McNeil v. Director, Patuxent Institution, 407 U. S. 245, 249-250 (1972); Jackson v. Indiana, 406 IT. S. 715, 738 (1972).
The Court of Appeals purported to be applying these principles in developing the first of its theories support*581ing a constitutional right to treatment. It first identified what it perceived to be the traditional bases for civil commitment — physical dangerousness to oneself or others, or a need for treatment — and stated:
''[W]here, as in Donaldson’s case, the rationale for confinement is the ‘parens patriae’ rationale that the patient is in need of treatment, the due process clause requires that minimally adequate treatment be in fact provided. . . . 'To deprive any citizen of his or her liberty upon the altruistic theory that the confinement is for humane therapeutic reasons and then fail to provide adequate treatment violates the very fundamentals of due process.’ ” 493 F. 2d, at 521.
The Court of Appeals did not explain its conclusion that the rationale for respondent’s commitment was that he needed treatment. The Florida statutes in effect during the period of his confinement did not require that a person who had been adjudicated incompetent and ordered committed either be provided with psychiatric treatment or released, and there was no such condition in respondent’s order of commitment. Cf. Rouse v. Cameron, 125 U. S. App. D. C. 366, 373 F. 2d 451 (1967). More important, the instructions which the Court of Appeals read as establishing an absolute constitutional right to treatment did not require the jury to make any findings regarding the specific reasons for respondent’s confinement or to focus upon any rights he may have had under state law. Thus, the premise of the Court of Appeals’ first theory must have been that, at least with respect to persons who are not physically dangerous, a State has no power to confine the mentally ill except for the purpose of providing them with treatment.
That proposition is surely not descriptive of the power traditionally exercised by the States in this area. *582Historically, and for a considerable period of time, subsidized custodial care in private foster homes or boarding houses was the most benign form of care provided incompetent or mentally ill persons for whom the States assumed responsibility. Until well into the 19th century the vast majority of such persons were simply restrained in poorhouses, almshouses, or jails. See A. Deutsch, The Mentally 111 in America 38-54, 11A-131 (2d ed. 1949). The few States that established institutions for the mentally ill during this early period were concerned primarily with providing a more humane place of confinement and only secondarily with “during” the persons sent there. See id., at 98-113.
As the trend toward state care of the mentally ill expanded, eventually leading to the present statutory schemes for protecting such persons, the dual functions of institutionalization continued to be recognized. While one of the goals of this movement was to provide medical treatment to those who could benefit from it, it was acknowledged that this could not be done in all cases and that there was a large range of mental illness for which no known “cure” existed. In time, providing places for the custodial confinement of the so-called “dependent insane” again emerged as the major goal of the States’ programs in this area and remained so well into this century. See id., at 228-271; D. Rothman, The Discovery of the Asylum 264-295 (1971).
In short, the idea that States may not confine the mentally ill except for the purpose of providing them with treatment is of very recent origin,4 and there is no historical basis for imposing such a limitation on state power. Analysis of the sources of the civil commitment power likewise lends no support to that notion. There can be little doubt that in the exercise of its police power *583a State may confine individuals solely to protect society from the dangers of significant antisocial acts or communicable disease. Cf. Minnesota ex rel. Pearson v. Probate Court, 309 U. S. 270 (1940); Jacobson v. Massachusetts, 197 U. S. 11, 25-29 (1905). Additionally, the States are vested with the historic parens patriae power, including the duty to protect “persons under legal disabilities to act for themselves.” Hawaii v. Standard Oil Co., 405 U. S. 251, 257 (1972). See also Mormon Church v. United States, 136 U. S. 1, 56-58 (1890). The classic example of this role is when a State undertakes to act as “ 'the general guardian of all infants, idiots, and lunatics.’ ” Hawaii v. Standard Oil Co., supra, at 257, quoting 3 W. Blackstone, Commentaries *47.
Of course, an inevitable consequence of exercising the parens patriae power is that the ward’s personal freedom will be substantially restrained, whether a guardian is appointed to control his property, he is placed in the custody of a private third party, or committed to an institution. Thus, however the power is implemented, due process requires that it not be invoked indiscriminately. At a minimum, a particular scheme for protection of the mentally ill must rest upon a legislative determination that it is compatible with the best interests of the affected class and that its members are unable to act for themselves. Cf. Mormon Church v. United States, supra. Moreover, the use of alternative forms of protection may be motivated by different considerations, and the justifications for one may not be invoked to rationalize another. Cf. Jackson v. Indiana, 406 U. S., at 737-738. See also American Bar Foundation, The Mentally Disabled and the Law 254-255 (S. Brakel & R. Rock ed. 1971).
However, the existence of some due process limitations on the parens patriae power does not justify the further conclusion that it may be exercised to confine a mentally *584ill person only if the purpose of the confinement is treatment. Despite many recent advances in medical knowledge, it remains a stubborn fact that there are many forms of mental illness which are not understood, some which are untreatable in the sense that no effective therapy has yet been discovered for them, and that rates of “cure” are generally low. See Schwitzgebel, The Right to Effective Mental Treatment, 62 Calif. L. Rev. 936, 941-948 (1974). There can be little responsible debate regarding “the uncertainty of diagnosis in this field and the tentativeness of professional judgment.” Greenwood v. United States, 350 U. S., at 375. See also Ennis & Litwack, Psychiatry and the Presumption of Expertise: Flipping Coins in the Courtroom, 62 Calif. L. Rev. 693, 697-719 (1974).5 Similarly, as previously observed, it is universally recognized as fundamental to effective therapy that the patient acknowledge his illness and cooperate with those attempting to give treatment; yet the failure of a large proportion of mentally ill persons to do so is a common phenomenon. See Katz, supra, 36 U. Chi. L. Rev., at 768-769. It may be that some persons in either of these categories,6 and there may be others, are unable to function in society and will suffer real harm to themselves unless provided with care in a sheltered environment. See, e. g., Lake v. Cameron, 124 U. S. App. D. C. *585264, 270-271, 364 F. 2d 657, 663-664 (1966) (dissenting opinion). At the very least, I am not able to say that a state legislature is powerless to make that kind of judgment. See Greenwood v. United States, supra.
B
Alternatively, it has been argued that, a Fourteenth Amendment right to treatment for involuntarily confined mental patients derives from the fact that many of the safeguards of the criminal process are not present in civil commitment. The Court of Appeals described this theory as follows:
a[A] due process right to treatment is based on the principle that when the three central limitations on the government's power to detain- — that detention be in retribution for a specific offense; that it be limited to a fixed term; and that it be permitted after a proceeding where the fundamental procedural safeguards are observed — are absent, there must be a quid pro quo extended by the government to justify confinement. And the quid pro quo most commonly recognized is the provision of rehabilitative treatment.” 493 F. 2d, at 522.
To the extent that this theory may be read to permit a State to confine an individual simply because it is willing to provide treatment, regardless of the subject's ability to function in society, it raises the gravest of constitutional problems, and I have no doubt the Court of Appeals would agree on this score. As a justification for a constitutional right to such treatment, the quid pro quo theory suffers from equally serious defects.
It is too well established to require extended discussion that due process is not an inflexible concept. Rather, its requirements are determined in particular instances by identifying and accommodating the inter*586ests of the individual and society. See, e. g., Morrissey v. Brewer, 408 U. S. 471, 480-484 (1972); McNeil v. Director, Patuxent Institution, 407 U. S., at 249-250; McKeiver v. Pennsylvania, 403 U. S. 528, 545-555 (1971) (plurality opinion). Where claims that the State is acting in the best interests of an individual are said to justify reduced procedural and substantive safeguards, this Court’s decisions require that they be “candidly appraised.” In re Gault, 387 U. S., at 21, 27-29. However, in so doing judges are not free to read their private notions of public policy or public health into the Constitution. Olsen v. Nebraska, 313 U. S. 236, 246-247 (1941).
The quid pro quo theory is a sharp departure from, and cannot coexist with, due process principles. As an initial matter, the theory presupposes that essentially the same interests are involved in every situation where a State seeks to confine an individual; that assumption, however, is incorrect. It is elementary that the justification for the criminal process and the unique deprivation of liberty which it can impose requires that it be invoked only for commission of a specific offense prohibited by legislative enactment. See Powell v. Texas, 392 U. S. 514, 541-544 (1968) (opinion of Black, J.).7 But it would be incongruous, for example, to apply the same limitation when quarantine is imposed by the State to protect the public from a highly communicable disease. See Jacobson v. Massachusetts, 197 U. S., at 29-30.
*587A more troublesome feature of the quid pro quo theory is that it would elevate a concern for essentially procedural safeguards into a new substantive constitutional right.8 Rather than inquiring whether strict standards of proof or periodic redetermination of a patient’s condition are required in civil confinement, the theory accepts the absence of such safeguards but insists that the State provide benefits which, in the view of a court, are adequate “compensation” for confinement. In light of the wide divergence of medical opinion regarding the diagnosis of and proper therapy for mental abnormalities, that prospect is especially troubling in this area and cannot be squared with the principle that “courts may not substitute for the judgments of legislators their own understanding of the public welfare, but must instead concern themselves with the validity under the Constitution of the methods which the legislature has selected.” In re Gault, 387 U. S., at 71 (Harlan, J., concurring and dissenting). Of course, questions regarding the adequacy of procedure and the power of a State to continue particular confinements are ultimately for the courts, aided by expert opinion to the extent that is found helpful. But I am not persuaded that we should abandon the traditional limitations on the scope of judicial review.
C
In sum, I cannot accept the reasoning of the Court of Appeals and can discern no basis for equating an involuntarily committed mental patient’s unquestioned constitutional right not to be confined without due proc*588ess of law with a constitutional right to treatment.9 Given the present state of medical knowledge regarding abnormal human behavior and its treatment, few things would be more fraught with peril than to irrevocably condition a State’s power to protect the mentally ill upon the providing of “such treatment as will give [them] a *589realistic opportunity to be cured.” Nor can I accept the theory that a State may lawfully confine an individual thought to need treatment and justify that deprivation of liberty solely by providing some treatment. Our concepts of due process would not tolerate such a “tradeoff.” Because the Court of Appeals' analysis could be read as authorizing those results, it should not be followed.
I have difficulty understanding how the issue of immunity can be resolved on this record and hence it is very likely a new trial on this issue may be required; if that is the case I would hope these sensitive and important issues would have the benefit of more effective presentation and articulation on behalf of petitioner.
The Court’s reference to “milieu therapy,” ante, at 569, may be construed as disparaging that concept. True, it is capable of being used simply to cloak official indifference, but the reality is that some mental abnormalities respond to no known treatment. Also, some mental patients respond, as do persons suffering from a variety of physiological ailments, to what is loosely called “milieu treatment,” i. e., keeping them comfortable, well nourished, and in a protected environment. It is not for us to say in the baffling field of psychiatry that “milieu therapy” is always a pretense.
That petitioner’s counsel failed to raise this issue is not a reason *580why it should not be considered with respect to immunity in light of the Court’s holding that the defense was preserved for appellate review.
See Editorial, A New Right, 46 A. B. A. J. 516 (1960).
Indeed, there is considerable debate concerning the threshold questions of what constitutes “mental disease” and “treatment.” See Szasz, The Right to Health, 57 Geo. L. J. 734 (1969).
Indeed, respondent may have shared both of these characteristics. His illness, paranoid schizophrenia, is notoriously unsusceptible to treatment, see Livermore, Malmquist, & Meehl, On the Justifications for Civil Commitment, 117 U. Pa. L. Rev. 75, 93, and n. 52 (1968), and the reports of the Florida State Hospital staff which were introduced into evidence expressed the view that he was unwilling to acknowledge his illness and was generally uncooperative.
This is not to imply that I accept all of the Court of Appeals’ conclusions regarding the limitations upon the States’ power to detain persons who commit crimes. For example, the notion that confinement must be “for a fixed term” is difficult to square with the widespread practice of indeterminate sentencing, at least where the upper limit is a life sentence.
Even advocates of a right to treatment have criticized the quid pro quo theory on this ground. E. g., Developments in the Law— Civil Commitment of the Mentally 111, 87 Harv. L. Rev. 1190, 1325 n. 39 (1974).
It should be pointed out that several issues which the Court has touched upon in other contexts are not involved here. As the Court’s opinion makes plain, this is not a case of a person’s seeking release because he has been confined “without ever obtaining a judicial determination that such confinement is warranted.” McNeil v. Director, Patuxent Institution, 407 U. S. 245, 249 (1972). Although respondent’s amended complaint alleged that his 1956 hearing before the Pinellas County Court was procedurally defective and ignored various factors relating to the necessity for commitment, the persons to whom those allegations applied were either not served with process or dismissed by the District Court prior to trial. Respondent has not sought review of the latter rulings, and this case does not involve the rights of a person in an initial competency or commitment proceeding. Cf. Jackson v. Indiana, 406 U. S. 715, 738 (1972); Specht v. Patterson, 386 U. S. 605 (1967); Minnesota ex rel. Pearson v. Probate Court, 309 U. S. 270 (1940).
Further, it was not alleged that respondent was singled out for discriminatory treatment by the staff of Florida State Hospital or that patients at that institution were denied privileges generally available to other persons under commitment in Florida. Thus, the question whether different bases for commitment justify differences in conditions of confinement is not involved in this litigation. Cf. Jackson v. Indiana, supra, at 723-730; Baxstrom v. Herold, 383 U. S. 107 (1966).
Finally, there was no evidence whatever that respondent was abused or mistreated at Florida State Hospital or that the failure to provide him with treatment aggravated his condition. There was testimony regarding the general quality of life at the hospital, but the jury was not asked to consider whether respondent’s confinement was in effect “punishment” for being mentally ill. The record provides no basis for concluding, therefore, that respondent was denied rights secured by the Eighth and Fourteenth Amendments. Cf. Robinson v. California, 370 U. S. 660 (1962).