dissenting.
The New York Family Court Act governs the treatment of persons between 7 and 16 years of age who are alleged to have committed acts that, if committed by adults, would *282constitute crimes.1 The Act contains two provisions that authorize the detention of juveniles arrested for offenses covered by the Act2 for up to 17 days pending adjudication of their guilt.3 Section 320.5(3)(a) empowers a judge of the New York Family Court to order detention of a juvenile if he finds “there is a substantial probability that [the juvenile] will not appear in court on the return date.” Section 320.5(3)(b), the provision at issue in these cases, authorizes detention if the judge finds “there is a serious risk [the juvenile] may before the return date commit an act which if committed by an adult would constitute a crime.”4
*283There are few limitations on § 320.5(3)(b). Detention need not be predicated on a finding that there is probable cause to believe the child committed the offense for which he was arrested. The provision applies to all juveniles, regardless of their prior records or the severity of the offenses of which they are accused. The provision is not limited to the prevention of dangerous crimes; a prediction that a juvenile if released may commit a minor misdemeanor is sufficient to justify his detention. Aside from the reference to “serious risk,” the requisite likelihood that the juvenile will misbehave before his trial is not specified by the statute.
The Court today holds that preventive detention of a juvenile pursuant to § 320.5(3)(b) does not violate the Due Process Clause. Two rulings are essential to the Court’s decision: that the provision promotes legitimate government objectives important enough to justify thé abridgment of the detained juveniles’ liberty interests, ante, at 274; and that the provision incorporates procedural safeguards sufficient to prevent unnecessary or arbitrary impairment of constitutionally protected rights, ante, at 277, 279-280. Because I disagree with both of those rulings, I dissent.
I
The District Court made detailed findings, which the Court of Appeals left undisturbed, regarding the manner in which § 320.5(3)(b) is applied in practice. Unless clearly erroneous, those findings are binding upon us, see Fed. Rule Civ. Proc. 52(a), and must guide our analysis of the constitutional questions presented by these cases.
The first step in the process that leads to detention under § 320.5(3)(b) is known as “probation intake.” A juvenile may arrive at intake by one of three routes: he may be brought there directly by an arresting officer; he may be detained for a brief period after his arrest and then taken to intake; he may be released upon arrest and directed to appear at a designated time. United States ex rel. Martin v. Strasburg, *284513 F. Supp. 691, 701 (SDNY 1981). The heart of the intake procedure is a 10-to-40-minute interview of the juvenile, the arresting officer, and sometimes the juvenile’s parent or guardian. The objectives of the probation officer conducting the interview are to determine the nature of the offense the child may have committed and to obtain some background information on him. Ibid.
On the basis of the information derived from the interview and from an examination of the juvenile’s record, the probation officer decides whether the case should be disposed of informally (“adjusted”) or whether it should be referred to the Family Court. If the latter, the officer makes an additional recommendation regarding whether the juvenile should be detained. “There do not appear to be any governing criteria which must be followed by the probation officer in choosing between proposing detention and parole . . . .” Ibid.
The actual decision whether to detain a juvenile under §320.5(3)(b) is made by a Family Court judge at what is called an “initial appearance” — a brief hearing resembling an arraignment.5 Id., at 702. The information on which the judge makes his determination is very limited. He has before him a “petition for delinquency” prepared by a state agency, charging the juvenile with an offense, accompanied with one or more affidavits attesting to the juvenile’s involvement. Ordinarily the judge has in addition the written report and recommendation of the probation officer. However, the probation officer who prepared the report rarely attends the hearing. Ibid. Nor is the complainant likely to appear. Consequently, “[ojften there is no one present with-personal knowledge of what happened.” Ibid.
In the typical case, the judge appoints counsel for the juvenile at the time his case is called. Thus, the lawyer has no opportunity to make an independent inquiry into the juvenile’s background or character, and has only a few minutes to *285prepare arguments on the child’s behalf. Id., at 702, 708. The judge ordinarily does not interview the juvenile, id., at 708, makes no inquiry into the truth of allegations in the petition, id., at 702, and does not determine whether there is probable cause to believe the juvenile committed the offense.6 The typical hearing lasts between 5 and 15 minutes, and the judge renders his decision immediately afterward. Ibid.
Neither the statute nor any other body of rules guides the efforts of the judge to determine whether a given juvenile is likely to commit a crime before his trial. In making detention decisions, “each judge must rely on his own subjective *286judgment, based on the limited information available to him at court intake and whatever personal standards he himself has developed in exercising his discretionary authority under the statute.” Ibid. Family Court judges are not provided information regarding the behavior of juveniles over whose cases they have presided, so a judge has no way of refining the standards he employs in making detention decisions. Id., at 712.
After examining a study of a sample of 34 cases in which juveniles were detained under §320.5(3)(b)7 along with various statistical studies of pretrial detention of juveniles in New York,8 the District Court made findings regarding the *287circumstances in which the provision habitually is invoked. Three of those findings are especially germane to appellees’ challenge to the statute. First, a substantial number of “first offenders” are detained pursuant to §320.5(3)(b). For example, at least 5 of the 34 juveniles in the sample had no prior contact with the Family Court before being detained and at least 16 had no prior adjudications of delinquency. Id., at 695-700.9 Second, many juveniles are released — for periods ranging from five days to several weeks — after their arrests and are then detained under §320.5(3)(b), despite the absence of any evidence of misconduct during the time between their arrests and “initial appearances.” Sixteen of the thirty-four cases in the sample fit this pattern. Id., at 705, 713-714. Third, “the overwhelming majority” of the juveniles detained under §320.5(3)(b) are released either before or immediately after their trials, either unconditionally or on parole. Id., at 705. At least 23 of the juveniles in the sample fell into this category. Martin v. Strasburg, 689 F. 2d 365, 369, n. 19 (CA2 1982); see 513 F. Supp., at 695-700.
Finally, the District Court made a few significant findings concerning the conditions associated with “secure detention” pursuant to § 320.5(3)(b).10 In a “secure facility,” “[t]he juveniles are subjected to strip-searches, wear institutional clothing and follow institutional regimen. At Spofford [Juvenile Detention Center], which is a secure facility, some juveniles who have had dispositional determinations and were awaiting *288placement (long term care) commingle with those in pretrial detention (short term care).” Id., at 695, n. 5.
It is against the backdrop of these findings that the contentions of the parties must be examined.
H
>
As the majority concedes, ante, at 263, the fact that §320.5(3)(b) applies only to juveniles does not insulate the provision from review under the Due Process Clause. “[Njeither the Fourteenth Amendment nor the Bill of Rights is for adults alone.” In re Gault, 387 U. S. 1, 13 (1967). Examination of the provision must of course be informed by a recognition that juveniles have different needs and capacities than adults, see McKeiver v. Pennsylvania, 403 U. S. 528, 550 (1971), but the provision still “must measure up to the essentials of due process and fair treatment,” Kent v. United States, 383 U. S. 541, 562 (1966).
To comport with “fundamental fairness,” §320.5(3)(b) must satisfy two requirements. First, it must advance goals com-' mensúrate with the burdens it imposes on constitutionally protected interests. Second, it must not punish the juveniles to whom it applies.
The majority only grudgingly and incompletely acknowledges the applicability of the first of these tests, but its grip on the cases before us is undeniable. It is manifest that §320.5(3)(b) impinges upon fundamental rights. If the “liberty” protected by the Due Process Clause means anything, it means freedom from physical restraint. Ingraham v. Wright, 430 U. S. 651, 673-674 (1977); Board of Regents v. Roth, 408 U. S. 564, 572 (1972).. Only a very important government interest can justify deprivation of liberty in this basic sense.11
*289The majority seeks to evade the force of this principle by discounting the impact on a child of incarceration pursuant to §320.5(3)(b). The curtailment of liberty consequent upon detention of a juvenile, the majority contends, is mitigated by the fact that “juveniles, unlike adults, are always in some form of custody.” Ante, at 265. In any event, the majority argues, the conditions of confinement associated with “secure detention” under §320.5(3)(b) are not unduly burdensome. Ante, at 271. These contentions enable the majority to suggest that §320.5(3)(b) need only advance a “legitimate state objective” to satisfy the strictures of the Due Process Clause. Ante, at 256-257, 263-264, 274.12
The majority’s arguments do not survive scrutiny. Its characterization of preventive detention as merely a transfer of custody from a parent or guardian to the State is difficult to take seriously. Surely there is a qualitative difference between imprisonment and the condition of being subject to *290the supervision and control of an adult who has one’s best interests at heart. And the majority’s depiction of the nature of confinement under §320.5(3)(b) is insupportable on this record. As noted above, the District Court found that secure detention entails incarceration in a facility closely resembling a jail and that pretrial detainees are sometimes mixed with juveniles who have been found to be delinquent. Supra, at 287-288. Evidence adduced at trial reinforces these findings. For example, Judge Quinones, a Family Court Judge with eight years of experience, described the conditions of detention as follows:
“Then again, Juvenile Center, as much as we might try, is not the most pleasant place in the world. If you put them in detention, you are liable to be exposing these youngsters to all sorts of things. They are liable to be exposed to assault, they are liable to be exposed to sexual assaults. You are taking the risk of putting them together with a youngster that might be much worse than they, possibly might be, and it might have a bad effect in that respect.” App. 270.
Many other observers of the circumstances of juvenile detention in New York have come to similar conclusions.13
*291In short, fairly viewed, pretrial detention of a juvenile pursuant to §320.5(3)(b) gives rise to injuries comparable to those associated with imprisonment of an adult. In both situations, the detainee suffers stigmatization and severe limitation of his freedom of movement. See In re Winship, 397 U. S. 358, 367 (1970); In re Gault, 387 U. S., at 27. Indeed, the impressionability of juveniles may make the experience of incarceration more injurious to them than to adults; all too quickly juveniles subjected to preventive detention come to see society at large as hostile and oppressive and to regard themselves as irremediably “delinquent.”14 Such serious injuries to presumptively innocent persons — encompassing the curtailment of their constitutional rights to liberty — can be justified only by a weighty public interest that is substantially advanced by the statute.15
The applicability of the second of the two tests is admitted even by the majority. In Bell v. Wolfish, 441 U. S. 520, 535 *292(1979), the Court held that an adult may not be punished prior to determination that he is guilty of a crime.16 The majority concedes, as it must, that this principle applies to juveniles. Ante, at 264, 269. Thus, if the only purpose substantially advanced by §320.5(3)(b) is punishment, the provision must be struck down.
For related reasons, §320.5(3)(b) cannot satisfy either of the requirements discussed above that together define “fundamental fairness” in the context of pretrial detention.
B
Appellants and the majority contend that §320.5(3)(b) advances a pair of intertwined government objectives: “protecting the community from crime,” ante, at 264, and “protecting a juvenile from the consequences of his criminal activity,” ante, at 266. More specifically, the majority argues that detaining a juvenile for a period of up to 17 days prior to his trial has two desirable effects: it protects society at large from the crimes he might have committed during that period if released; and it protects the juvenile himself “both from potential physical injury which may be suffered when a victim fights back or a policeman attempts to make an arrest and from the downward spiral of criminal activity into which peer pressure may lead the child.” Ante, at 264-266.
Appellees and some amici argue that public purposes of this sort can never justify incarceration of a person who has not been adjudicated guilty of a crime, at least in the absence of a determination that there exists probable cause to believe he committed a criminal offense.17 We need not reach that *293categorial argument in these cases because, even if the purposes identified by the majority are conceded to be compelling, they are not sufficiently promoted by detention pursuant to §320.5(3)(b) to justify the concomitant impairment of the juveniles’ liberty interests.18 To state the case more precisely, two circumstances in combination render §320.5(3)(b) invalid in toto: in the large majority of cases in which the provision is invoked, its asserted objectives are either not advanced at all or are only minimally promoted; and, as the provision is written and administered by the state courts, the cases in which its asserted ends are significantly advanced cannot practicably be distinguished from the cases in which they are not.
1
Both of the courts below concluded that only occasionally and accidentally does pretrial detention of a juvenile under §320.5(3)(b) prevent the commission of a crime. Three subsidiary findings undergird that conclusion. First, Family Court judges are incapable of determining which of the juveniles who appear before them would commit offenses before their trials if left at large and which would not. In part, this incapacity derives from the limitations of current knowledge concerning the dynamics of human behavior. On the basis of evidence adduced at trial, supplemented by a thorough review of the secondary literature, see 513 F. Supp., at 708-712, and nn. 31-32, the District Court found that “no diagnostic tools have as yet been devised which enable even the most highly trained criminologists to predict reliably which juveniles will engage in violent crime.” Id., at 708. The evidence supportive of this finding is overwhelm*294ing.19 An independent impediment to identification of the defendants who would misbehave if released is the paucity of data available at an initial appearance. The judge must make his decision whether to detain a juvenile on the basis of a set of allegations regarding the child’s alleged offense, a cursory review of his background and criminal record, and the recommendation of a probation officer who, in the typical case, has seen the child only once. Id., at 712. In view of this scarcity of relevant information, the District Court credited the testimony of appellees’ expert witness, who “stated that he would be surprised if recommendations based on intake interviews were better than chance and assessed the judge’s subjective prognosis about the probability of future crime as only 4% better than chance — virtually wholly unpredictable.” Id., at 708.20
*295Second, §320.5(3)(b) is not limited to classes of juveniles whose past conduct suggests that they are substantially more likely than average juveniles to misbehave in the immediate future. The provision authorizes the detention of persons arrested for trivial offenses21 and persons without any prior contacts with juvenile court. Even a finding that there is probable cause to believe a juvenile committed the offense with which he was charged is not a prerequisite to his detention. See supra, at 285, and n. 6.22
*296Third, the courts below concluded that circumstances surrounding most of the cases in which § 320.5(3)(b) has been invoked strongly suggest that the detainee would not have committed a crime during the period before his trial if he had been released. In a significant proportion of the cases, the juvenile had been released after his arrest and had not committed any reported crimes while at large, see supra, at 287; it is not apparent why a juvenile would be more likely to misbehave between his initial appearance and his trial than between his arrest and initial appearance. Even more telling is the fact that “the vast majority” of persons detained under §320.5(3)(b) are released either before or immediately after their trials. 698 F. 2d, at 369; see 513 F. Supp., at 705. The inference is powerful that most detainees, when examined more carefully than at their initial appearances, are deemed insufficiently dangerous to warrant further incarceration.23
The rarity with which invocation of §320.5(3)(b) results in detention of a juvenile who otherwise would have committed a crime fatally undercuts the two public purposes assigned to the statute by the State and the majority. The argument that §320.5(3)(b) serves “the State’s ‘parens patriae interest in preserving and promoting the welfare of the child,”’ ante, at 265 (citation omitted), now appears particularly hollow. Most juveniles detained pursuant to the provision are not *297benefited thereby, because they would not have committed crimes if left to their own devices (and thus would not have been exposed to the risk of physical injury or the perils of the cycle of recidivism, see ante, at 266). On the contrary, these juveniles suffer several serious harms: deprivation of liberty and stigmatization as “delinquent” or “dangerous,” as well as impairment of their ability to prepare their legal defenses.24 The benefits even to those few juveniles who would have committed crimes if released are not unalloyed; the gains to them are partially offset by the aforementioned injuries. In view of this configuration of benefits and harms, it is not surprising that Judge Quinones repudiated the suggestion that detention under §320.5(3)(b) serves the interests of the detainees. App. 269-270.
The argument that § 320.5(3)(b) protects the welfare of the community fares little better. Certainly the public reaps no benefit from incarceration of the majority of the detainees who would not have committed any crimes had they been released. Prevention of the minor offenses that would have been committed by a small proportion of the persons detained confers only a slight benefit on the community.25 Only in occasional cases does incarceration of a juvenile pending his trial serve to prevent a crime of violence and thereby significantly promote the public interest. Such an infrequent and haphazard gain is insufficient to justify curtailment of the lib*298erty interests of all the presumptively innocent juveniles who would have obeyed the law pending their trials had they been given the chance.26
2
The majority seeks to deflect appellees’ attack on the constitutionality of §320.5(3)(b) by contending that they have framed their argument too broadly. It is possible, the majority acknowledges, that “in some circumstances detention of a juvenile [pursuant to §320.5(3)(b)] would not pass constitutional muster. But the validity of those detentions must be determined on a case-by-case basis.” Ante, at 273; see ante, at 268-269, n. 18. The majority thus implies that, even if the Due Process Clause is violated by most detentions under §320.5(3)(b) because those detainees would not have committed crimes if released, the statute nevertheless is not invalid “on its face” because detention of those persons who would have committed a serious crime comports with the Constitution. Separation of the properly detained juveniles from the improperly detained juveniles must be achieved through “case-by-case” adjudication.
There are some obvious practical impediments to adoption of the majority’s proposal. Because a juvenile may not be incarcerated under §320.5(3)(b) for more than 17 days, it *299would be impracticable for a particular detainee to secure his freedom by challenging the constitutional basis of his detention; by the time the suit could be considered, it would have been rendered moot by the juvenile’s release or long-term detention pursuant to a delinquency adjudication.27 Nor could an individual detainee avoid the problem of mootness by filing a suit for damages or for injunctive relief. This Court’s declaration that §320.5(3)(b) is not unconstitutional on its face would almost certainly preclude a finding that detention of a juvenile pursuant to the statute violated any clearly established constitutional rights; in the absence of such a finding all state officials would be immune from liability in damages, see Harlow v. Fitzgerald, 457 U. S. 800 (1982). And, under current doctrine pertaining to the standing of an individual victim of allegedly unconstitutional conduct to obtain an injunction against repetition of that behavior, it is far from clear that an individual detainee would be able to obtain *300an equitable remedy. Compare INS v. Delgado, 466 U. S. 210, 217, n. 4 (1984), with Los Angeles v. Lyons, 461 U. S. 95, 105-106 (1983).
But even if these practical difficulties could be surmounted, the majority’s proposal would be inadequate. Precisely because of the unreliability of any determination whether a particular juvenile is likely to commit a crime between his arrest and trial, see swpra, at 293-294, no individual detainee would be able to demonstrate that he would have abided by the law had he been released. In other words, no configuration of circumstances would enable a juvenile to establish that he fell into the category of persons unconstitutionally detained rather than the category constitutionally detained.28 Thus, to protect the rights of the majority of juveniles whose incarceration advances no legitimate state interest, §320.5(3)(b) must be held unconstitutional “on its face.”
C
The findings reviewed in the preceding section lend credence to the conclusion reached by the courts below: §320.5(3)(b) “is utilized principally, not for preventive purposes, but to impose punishment for unadjudicated criminal acts.” 689 F. 2d, at 372; see 513 F. Supp., at 715-717.
The majority contends that, of the many factors we have considered in trying to determine whether a particular sanction constitutes “punishment,” see Kennedy v. Mendoza-Martinez, 372 U. S. 144, 168-169 (1963), the most useful are “whether an alternative purpose to which [the sanction] may *301rationally be connected is assignable for it, and whether it appears excessive in relation to the alternative purpose assigned,” ibid, (footnotes omitted). See ante, at 269. Assuming, arguendo, that this test is appropriate, but cf. Bell v. Wolfish, 441 U. S., at 564-565 (Marshall, J., dissenting),. it requires affirmance in these cases. The alternative purpose assigned by the State to §320.5(3)(b) is the prevention of crime by the detained juveniles. But, as has been shown, that objective is advanced at best sporadically by the provision. Moreover, §320.5(3)(b) frequently is invoked under circumstances in which it is extremely unlikely that the juvenile in question would commit a crime while awaiting trial. The most striking of these cases involve juveniles who have been at large without mishap for a substantial period of time prior to their initial appearances, see supra, at 287, and detainees who are adjudged delinquent and are nevertheless released into the community. In short, §320.5(3)(b) as administered by the New York courts surely “appears excessive in relation to” the putatively legitimate objectives assigned to it.
The inference that § 320.5(3)(b) is punitive in nature is supported by additional materials in the record. For example, Judge Quinones and even appellants’ counsel acknowledged that one of the reasons juveniles detained pursuant to §320.5(3)(b) usually are released after the determination of their guilt is that the judge decides that their pretrial detention constitutes sufficient punishment. 689 F. 2d, at 370-371, and nn. 27-28. Another Family Court Judge-admitted using “preventive detention” to punish one of the juveniles in the sample. 513 F. Supp., at 708.29
*302In summary, application of the litmus test the Court recently has used to identify punitive sanctions supports the finding of the lower courts that preventive detention under §320.5(3)(b) constitutes punishment. Because punishment of juveniles before adjudication of their guilt violates the Due Process Clause, see supra, at 291-292, the provision cannot stand.
I — I h — I I — I
If the record did not establish the impossibility, on the basis of the evidence available to a Family Court judge at a §320.5(3)(b) hearing, of reliably predicting whether a given juvenile would commit a crime before his trial, and if the purposes relied upon by the State were promoted sufficiently to justify the deprivations of liberty effected by the provision, I would nevertheless still strike down §320.5(3)(b) because of the absence of procedural safeguards in the provision. As Judge Newman, concurring in the Court of Appeals observed, “New York’s statute is unconstitutional because it permits liberty to be denied, prior to adjudication of guilt, in the exercise of unfettered discretion as to an issue of considerable uncertainty — likelihood of future criminal behavior.” 689 F. 2d, at 375.
Appellees point out that § 320.5(3)(b) lacks two crucial procedural constraints. First, a New York Family Court judge is given no guidance regarding what kinds of evidence he should consider or what weight he should accord different sorts of material in deciding whether to detain a juvenile.30 For example, there is no requirement in the statute that the *303judge take into account the juvenile’s background or current living situation. Nor is a judge obliged to attach significance to the nature of a juvenile’s criminal record or the severity of the crime for which he was arrested.31 Second, § 320.5(3)(b) does not specify how likely it must be that a juvenile will commit a crime before his trial to warrant his detention. The provision indicates only that there must be a “serious risk” that he will commit an offense and does not prescribe the standard of proof that should govern the judge’s determination of that issue.32
Not surprisingly, in view of the lack of directions provided by the statute, different judges have adopted different ways of estimating the chances whether a juvenile will misbehave in the near future. “Each judge follows his own individual approach to [the detention] determination.” 513 F. Supp., at 702; see App. 265 (testimony of Judge Quinones). This discretion exercised by Family Court judges in making detention decisions gives rise to two related constitutional problems. First, it creates an excessive risk that juveniles will be detained “erroneously” — i. e., under circumstances in which no public interest would be served by their incarceration. Second, it fosters arbitrariness and inequality in a decisionmaking process that impinges upon fundamental rights.
A
One of the purposes of imposing procedural constraints on decisions affecting life, liberty, or property is to reduce the *304incidence of error. See Fuentes v. Shevin, 407 U. S. 67, 80-81 (1972). In Mathews v. Eldridge, 424 U. S. 319 (1976), the Court identified a complex of considerations that has proved helpful in determining what protections are constitutionally required in particular contexts to achieve that end:
“[identification of the specific dictates of due process generally requires consideration of three distinct factors: First, the private interest that will be affected by the official action; second, the risk of an erroneous deprivation of such interest through the procedures used, and the probable value, if any, of additional or substitute procedural safeguards; and finally, the Government’s interest, including the function involved and the fiscal and administrative burdens that the additional or substitute procedural requirement would entail.” Id., at 335.
As Judge Newman recognized, 689 F. 2d, at 375-376, a review of these three factors in the context of New York’s preventive-detention scheme compels the conclusion that the Due Process Clause is violated by §320.5(3)(b) in its present form. First, the private interest affected by a decision to detain a juvenile is personal liberty. Unnecessary abridgment of such a fundamental right, see supra, at 288, should be avoided if at all possible.
Second, there can be no dispute that there is a serious risk under the present statute that a juvenile will be detained erroneously — i. e., despite the fact that he would not commit a crime if released. The findings of fact reviewed in the preceding sections make it apparent that the vast majority of detentions pursuant to §320.5(3)(b) advance no state interest; only rarely does the statute operate to prevent crime. See supra, at 297-298. This high incidence of demonstrated error should induce a reviewing court to exercise utmost care in ensuring that no procedures could be devised that would improve the accuracy of the decisionmaking process. Opportunities for improvement in the extant regime are apparent *305even to a casual observer. Most obviously, some measure of guidance to Family Court judges regarding the evidence they should consider and the standard of proof they should use in making their determinations would, surely contribute to the quality of their detention determinations.33
The majority purports to see no value in such additional safeguards, contending that activity of estimating the likelihood that a given juvenile will commit a crime in the near future involves subtle assessment of a host of variables, the precise weight of which cannot be determined in advance. Ante, at 279. A review of the hearings that resulted in the detention of the juveniles included in the sample of 34 cases reveals the majority’s depiction of the decisionmaking process to be hopelessly idealized. For example, the operative portion of the initial appearance of Tyrone Parson, the three-card monte player,34 consisted of the following:
“COURT OFFICER: Will you identify yourself.
“TYRONE PARSON: Tyrone Parson, Age 15.
“THE COURT: Miss Brown, how many times has
Tyrone been known to the Court?
“MISS BROWN: Seven times.
*306“THE COURT: Remand the respondent.” Petitioners’ Exhibit 18a.35
This kind of parody of reasoned decisionmaking would be less likely to occur if judges were given more specific and mandatory instructions regarding the information they should consider and the manner in which they should assess it.
Third and finally, the imposition of such constraints on the deliberations of the Family Court judges would have no adverse effect on the State’s interest in detaining dangerous juveniles and would give rise to insubstantial administrative burdens. For example, a simple directive to Family Court judges to state on the record the significance they give to the seriousness of the offense of which a juvenile is accused and to the nature of the juvenile’s background would contribute materially to the quality of the decisionmaking process without significantly increasing the duration of initial appearances.
In summary, the three factors enumerated in Mathews in combination incline overwhelmingly in favor of imposition of more stringent constraints on detention determinations under § 320.5(3)(b). Especially in view of the impracticability of correcting erroneous decisions through judicial review, see supra, at 298-300, the absence of meaningful procedural safeguards in the provision renders it invalid. See Santosky v. Kramer, 455 U. S. 745, 757, and n. 9 (1982).
B
A principle underlying many of our prior decisions in various doctrinal settings is that government officials may not be accorded unfettered discretion in making decisions that *307impinge upon fundamental rights. Two concerns underlie this principle: excessive discretion fosters inequality in the distribution of entitlements and harms, inequality which is especially troublesome when those benefits and burdens are great; and discretion can mask the use by officials of illegitimate criteria in allocating important goods and rights.
So, in striking down on vagueness grounds a vagrancy ordinance, we emphasized the “unfettered discretion it places in the hands of the . . . police.” Papachristou v. City of Jacksonville, 405 U. S. 156, 168 (1972). Such flexibility was deemed constitutionally offensive because it “permits and encourages an arbitrary and discriminatory enforcement of the law.” Id., at 170. Partly for similar reasons, we have consistently held violative of the First Amendment ordinances which make the ability to engage in constitutionally protected speech “contingent upon the uncontrolled will of an official— as by requiring a permit or license which may be granted or withheld in the discretion of such official.” Staub v. City of Baxley, 355 U. S. 313, 322 (1958); accord, Shuttlesworth v. City of Birmingham, 394 U. S. 147, 151, 153 (1969). Analogous considerations inform our understanding of the dictates of the Due Process Clause. Concurring in the judgment in Zablocki v. Redhail, 434 U. S. 374 (1978), striking down a statute that conditioned the right to marry upon the satisfaction of child-support obligations, Justice Powell aptly observed:
“Quite apart from any impact on the truly indigent, the statute appears to ‘confer upon [the judge] a license for arbitrary procedure,’ in the determination of whether an applicant’s children are ‘likely thereafter to become public charges.’ A serious question of procedural due process is raised by this feature of standardless discretion, particularly in light of the hazards of prediction in this area.” Id., at 402, n. 4 (quoting Kent v. United States, 383 U. S., at 553).
*308The concerns that powered these decisions are strongly implicated by New York’s preventive-detention scheme. The effect of the lack of procedural safeguards constraining detention decisions under § 320.5(3)(b) is that the liberty of a juvenile arrested even for a petty crime is dependent upon the “caprice” of a Family Court judge. See 513 F. Supp., at 707. The absence of meaningful guidelines creates opportunities for judges to use illegitimate criteria when deciding whether juveniles should be incarcerated pending their trials — for example, to detain children for the express purpose of punishing them.36 Even the judges who strive conscientiously to apply the law have little choice but to assess juveniles’ dangerousness on the basis of whatever standards they deem appropriate.37 The resultant variation in detention decisions gives rise to a level of inequality in the deprivation of a fundamental right too great to be countenanced under the Constitution.
IV
The majority acknowledges — indeed, founds much of its argument upon — the principle that a State has both the power and the responsibility to protect the interests of the children within its jurisdiction. See Santosky v. Kramer, supra, at 766. Yet the majority today upholds a statute whose net impact on the juveniles who come within its purview is overwhelmingly detrimental. Most persons detained under the provision reap no benefit and suffer serious injuries thereby. The welfare of only a minority of the detainees is even arguably enhanced. The inequity of this regime, combined with *309the arbitrariness with which it is administered, is bound to disillusion its victims regarding the virtues of our system of criminal justice. I can see — and the majority has pointed to — no public purpose advanced by the statute sufficient to justify the harm it works.
I respectfully dissent.
N. Y. Jud. Law §§301.2(1), 302.1(1) (McKinney 1983) (hereinafter Family Court Act or FCA). Children aged 13 or over accused of murder and children aged 14 or over accused of kidnaping, arson, rape, or a few other serious crimes are exempted from the coverage of the Act and instead are prosecuted as “juvenile offenders” in the adult criminal courts. N. Y. Penal Law §§ 10.00(18), 30.00(2) (McKinney Supp. 1983-1984). For the sake of simplicity, offenses covered by the Family Court Act, as well as the more serious offenses enumerated above, hereinafter will be referred to generically as crimes.
Ironically, juveniles arrested for very serious offenses, see n. 1, supra, are not subject to preventive detention under this or any other provision.
Strictly speaking, “guilt” is never adjudicated under the Act; nor is the juvenile ever given a trial. Rather, whether the juvenile committed the offense is ascertained in a “factfinding hearing.” In most respects, however, such a hearing is the functional equivalent of an ordinary criminal trial. For example, the juvenile is entitled to counsel and the State bears the burden of demonstrating beyond a reasonable doubt that the juvenile committed the offense of which he is accused. See FCA §§341.2(1), 342.2(2); cf. In re Winship, 397 U. S. 358 (1970); In re Gault, 387 U. S. 1 (1967) (establishing constitutional limitations on the form of such proceedings in recognition of the severity of their impact upon juveniles). For convenience, the ensuing discussion will use the terminology associated with adult criminal proceedings when describing the treatment of juveniles in New York.
At the time appellees first brought their suit, the pertinent portions of FCA § 320.5(3) were embodied in FCA § 739(a). I agree with the majority that the reenactment of the crucial provision under a different numerical heading does not render the case moot. See ante, at 256, n. 2.
If the juvenile is detained upon arrest, this hearing must be held on the next court day or within 72 hours, whichever comes first. FCA § 307.3(4).
The majority admits that “the Family Court judge is not required to make a finding of probable cause at the initial appearance,” but contends that the juvenile has the option to challenge the sufficiency of the petition for delinquency on the ground that it fails to establish probable cause. Ante, at 276. None of the courts that have considered the constitutionality of New York’s preventive-detention system has suggested that a juvenile has a statutory right to a probable-cause determination before he is detained. The provisions cited by the majority for its novel reading of the statute provide only shaky support for its contention. FCA § 315.1, which empowers the juvenile to move to dismiss a petition lacking allegations sufficient to satisfy § 311.2, provides that “[a] motion to dismiss under this section must be made within the time provided for in section 332.2.” Section 332.2, in turn, provides that pretrial motions shall be made within 30 days after the initial appearance and before the factfinding hearing. If the juvenile has been detained, the judge is instructed to “hear and determine pre-trial motions on an expedited basis,” §332.2(4), but is not required to rule upon such motions peremptorily. In sum, the statutory scheme seems to contemplate that a motion to dismiss a petition for lack of probable cause, accompanied with “supporting affidavits, exhibits and memo-randa of law,” § 332.2(2), would be filed sometime after the juvenile is detained under § 320.5(3)(b). And there is no reason to expect that the ruling on such a motion would be rendered before the juvenile would in any event be entitled to a probable-cause hearing under § 325.1(2). That counsel for a juvenile ordinarily is not even appointed until a few minutes prior to the initial appearance, see supra, at 284 and this page, confirms this interpretation. The lesson of this foray into the tangled provisions of the New York Family Court Act is that the majority ought to adhere to our usual policy of relying whenever possible for interpretation of a state statute upon courts better acquainted with its terms and applications.
The majority refuses to consider the circumstances of these 34 cases, dismissing them as unrepresentative, ante, at 272, n. 21, and focuses instead on the lurid facts associated with the cases of the three named appel-lees. I cannot agree that the sample is entitled to so little weight. There was uncontested testimony at trial to the effect that the 34 cases were typical. App. 128 (testimony of Steven Hiltz, an attorney with 8V2 years of experience before the Family Court). At no point in this litigation have appellants offered an alternative selection of instances in which §320.5(3)(b) has been invoked. And most importantly, despite the fact that the District Court relied heavily on the sample when assessing the manner in which the statute is applied, see 513 F. Supp., at 695-700, appellants did not dispute before the Court of Appeals the representativeness of the 34 cases, see Martin v. Strasburg, 689 F. 2d 365, 369, n. 19 (CA2 1982). When the defendants in a plaintiff class action challenge on appeal neither the certification of the class, see ante, at 261, n. 10, nor the plaintiffs’ depiction of the character of the class, we ought to analyze the case as it comes to us and not try to construct a new version of the facts on the basis of an independent and selective review of the record.
As the Court of Appeals acknowledged, 689 F. 2d, at 369, n. 18, there are defects in all of the available statistical studies. Most importantly, none of the studies distinguishes persons detained under § 320.5(3)(a) from persons detained under §320.5(3)(b). However, these flaws did not disable the courts below from making meaningful — albeit rough — generalizations regarding the incidence of detention under the latter provision. Especially when conjoined with the sample of 34 cases submitted by appel-lees, see n. 7, supra, the studies are sufficient to support the three findings enumerated in the text. Even the majority, though it chastises appellees for failing to assemble better data, ante, at 272, and n. 21, does not suggest that those findings are clearly erroneous.
The figures in the text are taken from the District Court’s summary of the 34 cases in the sample. Review of the transcripts of the hearings in those cases reveals the actual number to be 9 and 23, respectively. See Petitioners’ Exhibits 6a, 11a, 12a, 14a, 15a, 16a, 19a, 24a, 35a.
The state director of detention services testified that, in 1978, approximately six times as many juveniles were ádmitted to “secure facilities” as to “non-secure facilities.” See 513 F. Supp., at 703, n. 8. These figures are not broken down as to persons detained under § 320.5(3)(a) and persons detained under § 320.5(3)(b). There seems no dispute, however, that most of the juveniles held under the latter provision are subjected to “secure detention.”
This principle underlies prior decisions of the Court involving various constitutional provisions as they relate to pretrial detention. In Gerstein *289v. Pugh, 420 U. S. 103, 113-114 (1975), we relied in part on the severity of “[t]he consequences of prolonged detention” in construing the Fourth Amendment to forbid pretrial incarceration of a suspect for an extended period of time without “a judicial determination of probable cause.” In Stack v. Boyle, 342 U. S. 1, 4-5 (1951), we stressed the importance of a person’s right to freedom until proved guilty in construing the Eighth Amendment to proscribe the setting of bail “at a figure higher than an amount reasonably calculated to” assure the presence of the accused at trial. Cf. Baker v. McCollan, 443 U. S. 137, 149-150, 153 (1979) (Stevens, J., dissenting).
The phrase “legitimate governmental objective” appears at several points in the opinion of the Court in Bell v. Wolfish, 441 U. S. 520 (1979), e. g., id., at 538-539, and the majority may be relying implicitly on that decision for the standard it applies in these cases. If so, the reliance is misplaced. Wolfish was exclusively concerned with the constitutionality of conditions of pretrial incarceration under circumstances in which the legitimacy of the incarceration itself was undisputed; the Court avoided any discussion of the showing a State must make in order to justify pretrial detention in the first instance. See id., at 533-534, and n. 15. The standard employed by the Court in Wolfish thus has no bearing on the problem before us.
All of the 34 juveniles in the sample were detained in Spofford Juvenile Center, the detention facility for New York City. Numerous studies of that facility have attested to its unsavory characteristics. See, e. g., Citizens’ Committee for Children of New York, Inc., Juvenile Detention Problems in New York City 3-4 (1970); J. Stone, R. Ruskin, & D. Goff, An Inquiry into the Juvenile Centers Operated by the Office of Probation 25-27, 52-54, 79-80 (1971). Conditions in Spofford have been successfully challenged on constitutional grounds (by a group of inmates of a different type), see Martarella v. Kelley, 359 P. Supp. 478 (SDNY 1973), but nevertheless remain grim, see Mayor’s Task Force on Spofford: First Report v, viii-ix, 20-21 (June 1978). Not surprisingly, a former New York City Deputy Mayor for Criminal Justice has averred that “Spofford is, in many ways, indistinguishable from a prison.” Petitioners’ Exhibit 30, ¶ 6 (affidavit of Herbert Sturz, June 29, 1978).
Cf. Aubry, The Nature, Scope and Significance of Pre-Trial Detention of Juveniles in California, 1 Black L. J. 160, 164 (1971).
This standard might be refined in one of two ways. First, it might be argued that, because § 320.5(3)(b) impinges upon “[ljiberty from bodily restraint,” which has long been “recognized as the core of the liberty protected by the Due Process Clause,” Greenholtz v. Nebraska Penal Inmates, 442 U. S. 1, 18 (1979) (Powell, J., concurring in part and dissenting in part), the provision can pass constitutional muster only if it promotes a “compelling” government interest. See People ex rel. Wayburn v. Schupf, 39 N. Y. 2d 682, 687, 350 N. E. 2d 906, 908 (1976) (requiring a showing of a “compelling State interest” to uphold § 320.5(3)(b)); cf. Shapiro v. Thompson, 394 U. S. 618, 634 (1969). Alternatively, it might be argued that the comparatively brief period of incarceration permissible under the provision warrants a slight lowering of the constitutional bar. Applying the principle that the strength of the state interest needed to legitimate a statute depends upon the degree to which the statute encroaches upon fundamental rights, see Williams v. Illinois, 399 U. S. 235, 259-260, 262-263 (1970) (Harlan, J., concurring in result), it might be held that an important — but not quite “compelling” — objective is necessary to sustain § 320.5(3)(b). In the present context, there is no need to choose between these doctrinal options, because §320.5(3)(b) would fail either test.
See also Ingraham v. Wright, 430 U. S. 651, 671-672, and n. 40, 673-674 (1977); Gregory v. Chicago, 394 U. S. Ill, 112 (1969); Thompson v. Louisville, 362 U. S. 199, 206 (1960).
Cf. Sellers v. United States, 89 S. Ct. 36, 38, 21 L. Ed. 2d 64, 67 (1968) (Black, J., in chambers) (questioning whether a defendant’s dangerousness can ever justify denial of bail).
An additional reason for not reaching appellees’ categorical objection to the purposes relied upon by the State is that the Court of Appeals did not pass upon the validity of those objectives. See 689 F. 2d, at 372. We are generally chary of deciding important constitutional questions not reached by a lower court.
See, e. g., American Psychiatric Association, Clinical Aspects of the Violent Individual 27-28 (1974); Cocozza & Steadman, The Failure of Psychiatric Predictions of Dangerousness: Clear and Convincing Evidence, 29 Rutgers L. Rev. 1084,1094-1101 (1976); Diamond, The Psychiatric Prediction of Dangerousness, 123 U. Pa. L. Rev. 439 (1974); Ennis & Litwack, Psychiatry and the Presumption of Expertise: Flipping Coins In the Courtroom, 62 Calif. L. Rev. 693 (1974); Schlesinger, The Prediction of Dangerousness in Juveniles: A Replication, 24 Crime & Delinquency 40, 47 (1978); Steadman & Cocozza, Psychiatry, Dangerousness and the Repetitively Violent Offender, 69 J. Crim. L. & C. 226, 229-231 (1978); Wenk, Robi-son, & Smith, Can Violence Be Predicted?, 18 Crime & Delinquency 393, 401 (1972); Preventive Detention: An Empirical Analysis, 6 Harv. Civ. Rights — Civ. Lib. L. Rev. 289 (1971).
The majority brushes aside the District Court’s findings on this issue with the remark that “a prediction of future criminal conduct. . . forms an important element in many decisions, and we have specifically rejected the contention . . . ‘that it is impossible to predict future behavior and that the question is so vague as to be meaningless.’” Ante, at 278-279 (footnote and citation omitted). Whatever the merits of the decisions upon which the majority relies, but cf., e. g., Barefoot v. Estelle, 463 U. S. 880, 909 (1983) (Marshall, J., dissenting), they do not control the problem before us. In each of the cases in which the Court has countenanced reliance upon a prediction of future conduct in a decisionmaking process impinging upon life or liberty, the affected person had already been convicted of a crime. See Greenholtz v. Nebraska Penal Inmates, 442 U. S. 1 (1979) *295(grant of parole); Jurek v. Texas, 428 U. S. 262 (1976) (death sentence); Morrissey v. Brewer, 408 U. S. 471 (1972) (parole revocation). The constitutional limitations upon the kinds of factors that may be relied on in making such decisions are significantly looser than those upon decisionmak-ing processes that abridge the liberty of presumptively innocent persons. Cf. United States v. Tucker, 404 U. S. 443, 446 (1972) (“[A] trial judge in the federal judicial system generally has wide discretion in determining what sentence to impose. . . . [Bjefore making that determination, a judge may appropriately conduct an inquiry broad in scope, largely unlimited either as to the kind of information he may consider, or the source from which it may come”).
For example, Tyrone Parson, aged 15, one of the members of the sample, was arrested for enticing others to play three-card monte. Petitioners’ Exhibit 18b. After being detained for five days under §320.5(3)(b), the petition against him was dismissed on the ground that “the offense alleged did not come within the provisions of the penal law.” 513 F. Supp., at 698-699.
In contrast to the breadth of the coverage of the Family Court Act, the District of Columbia adult preventive-detention statute that was upheld in United States v. Edwards, 430 A. 2d 1321 (D. C. 1981), cert. denied, 455 U. S. 1022 (1982), authorizes detention only of persons charged with one of a prescribed set of “dangerous crime[s]” or “crime[s] of violence.” D. C. Code §§ 23-1322(a)(l), (2) (1981).
Prediction whether a given person will commit a crime in the future is especially difficult when he has committed only minor crimes in the past. Cf. Baldosar v. Illinois, 446 U. S. 222, 231 (1980) (Powell, J., dissenting) (“No court can predict with confidence whether a misdemeanor defendant is likely to become a recidivist”).
By contrast, under the District of Columbia statute, see n. 21, supra, the judge is obliged before ordering detention to find, inter alia, a “substantial probability” that the defendant committed the serious crime for which he was arrested. D. C. Code § 23 — 1322(b)(2)(C) (1981).
Both courts below made this inference. See 689 F. 2d, at 372; 513 F. Supp., at 705. Indeed, the New York Court of Appeals, in upholding the statute, did not disagree with this explanation of the incidence of its application. People ex rel. Wayburn v. Schupf, 39 N. Y. 2d, at 690, 350 N. E. 2d, at 910.
Release (before or after trial) of some of the juveniles detained under § 320.5(3)(b) may well be due to a different factor: the evidence against them may be insufficient to support a finding of guilt. It is conceivable that some of those persons are so crime-prone that they would have committed an offense if not detained. But even the majority does not suggest that persons who could not be convicted of any crimes may nevertheless be imprisoned for the protection of themselves and the public.
See testimony of Steven Hiltz, App. 130-134 (describing the detrimental effects of pretrial detention of a juvenile upon the preparation and presentation of his defense); cf. Barker v. Wingo, 407 U. S. 514, 533 (1972); Bitter v. United States, 389 U. S. 15, 16-17 (1967) (per curiam); Stack v. Boyle, 342 U. S., at 8; Miller, Preventive Detention — A Guide to the Eradication of Individual Rights, 16 How. L. J. 1, 15 (1970).
Cf. Tribe, An Ounce of Detention: Preventive Justice in the World of John Mitchell, 56 Va. L. Rev. 371, 381 (1970) (“[Under a statute proposed by the Attorney General,] trivial property offenses may be deemed sufficiently threatening to warrant preventive imprisonment. No tenable concept of due process could condone a balance that gives so little weight to the accused’s interest in pretrial liberty”).
Some amici contend that a preventive-detention statute that, unlike §320.5(3)(b), covered only specific categories of juveniles and embodied stringent procedural safeguards would result in incarceration only of juveniles very likely to commit crimes of violence in the near future. E. g., Brief for American Bar Association as Amicus Curiae 9-14. It could be argued that, even though such a statute would unavoidably result in detention of some juveniles who would not have committed any offenses if released (because of the impossibility of reliably predicting the behavior of individual persons, see supra, at 293-294), the gains consequent upon the detention of the large proportion who would have committed crimes would be sufficient to justify the injuries to the other detainees. To decide the cases before us, we need not consider either the feasibility of such a scheme or its constitutionality.
The District Court, whose knowledge of New York procedural law surely exceeds ours, concluded that “[t]he short span of pretrial detention makes effective review impossible.” 513 F. Supp., at 708, n. 29. The majority dismisses this finding, along with a comparable finding by the Court of Appeals, see 689 F. 2d, at 373, as “mistaken.” Ante, at 280. But neither of the circumstances relied upon by the majority supports its confident judgment on this point. That the New York courts suspended their usual rules of mootness in order to consider an attack on the constitutionality of the statute as a whole, see People ex rel. Wayburn v. Schupf, 39 N. Y. 2d, at 686, 350 N. E. 2d, at 907-908, in no way suggests that they would be willing to do so if an individual detainee challenged the constitutionality of § 320.5(3)(b) as applied to him. The majority cites one case in which a detainee did obtain his release by securing a writ of habeas corpus. However, that case involved a juvenile who was not given a probable-cause hearing within six days of his detention — a patent violation of the state statute. See 513 F. Supp., at 708. That a writ of habeas corpus could be obtained on short notice to remedy a glaring statutory violation provides no support for the majority’s suggestion that individual detainees could effectively petition for release by challenging the constitutionality of their detentions.
This problem is exacerbated by the fact that Family Court judges, when making findings justifying a detention pursuant to § 320.5(3)(b), do not specify whether there is a risk that the juvenile would commit a serious crime or whether there is a risk that he would commit a petty offense. A finding of the latter sort should not be sufficient under the Due Process Clause to justify a juvenile’s detention. See supra, at 297-298, and n. 25. But a particular detainee has no way of ascertaining the grounds for his incarceration.
See transcript of the initial appearance of Ramon Ramos, #1356/80, Judge Heller presiding, Petitioners’ Exhibit 42, p. 11:
“This business now of being able to get guns, is now completely out of proportion. We are living in a jungle. We are living in a jungle, and it is time that these youths that are brought before the Court, know that they *302are in a Court, and that if these allegations are true, that they are going to pay the penalty.
“As for the reasons I just statefd] on the record, ... I am remand[ing] the respondent to the Commissioner of Juvenile Justice, secure detention.”
The absence of any limitations on the sorts of reasons that may support a determination that a child is likely to commit a crime if released means that the statutory requirement that the judge state “reasons” on the record, see ante, at 276, does not meaningfully constrain the decision-making process.
See 513 F. Supp., at 713:
“Whether the juvenile was a first offender with no prior conduct, whether the court was advised that the juvenile was an obedient son or was needed at home, whether probation intake recommended parole, the case histories in this record disclose that it was not unusual for the court to discount these considerations and order remand based on a 5 to 15 minute evaluation.”
Cf. Addington v. Texas, 441 U. S. 418, 431-433 (1979) (“clear and convincing” proof constitutionally required to justify civil commitment to mental hospital).
Judge Newman, concurring below, pointed to three other protections lacking in § 320.5(3)(b): “the statute places no limits on the crimes for which the person subject to detention has been arrested . . . , the judge ordering detention is not required to make any evaluation of the degree of likelihood that the person committed the crime of which he is accused[,]. . . [and] the statute places no limits on the type of crimes that the judge believes the detained juvenile might commit if released.” 689 F. 2d, at 377. In my view, the absence of these constraints is most relevant to the question whether the ends served by the statute can justify its broad reach, see Part II-B, supra. However, as Judge Newman observed, they could also be considered procedural flaws. Certainly, a narrowing of the categories of persons covered by § 320.5(3)(b), along the lines sketched by Judge Newman, would reduce the incidence of error in the application of the provision.
See n. 21, supra.
Parson’s case is not unique. The hearings accorded Juan Santiago and Daniel Nelson, for example, though somewhat longer in duration, were nearly as cavalier and undiscriminating. See Petitioners’ Exhibits 13a, 22a.
See n. 29, supra.
See 513 F. Supp., at 708:
“It is clear that the judge decides on pretrial detention for a variety of reasons — as a means of protecting the community, as the policy of the judge to remand, as an express punitive device, or because of the serious nature of the chargef,] among others” (citations omitted).