with whom Mr. Justice Brennan joins, dissenting.
There was a time, not so very long ago, when prisoners were regarded as “slave [s] of the State,” having “not only forfeited [their] liberty, but all [their] personal rights . . . Ruffin v. Commonwealth, 62 Va. 790, 796 (1871). In recent years, however, the courts increasingly have rejected this view, and with it the corollary which holds that courts should keep their “hands off” penal institutions.1 Today, however, the Court, in apparent fear of a prison reform organization that has the temerity to call itself a “union,” takes a giant step backwards toward that discredited conception of prisoners’ rights and the role of the courts. I decline to join in what I hope will prove to be a temporary retreat.
I
In Procunier v. Martinez, 416 U. S. 396 (1974), I set forth at some length my understanding of the First Amendment rights of prison inmates. The fundamental tenet I advanced is simply stated: “A prisoner does not shed . . . basic First Amendment rights at the prison gate. Rather, he ‘retains all the rights of an ordinary citizen except those expressly, or by necessary implication, taken from him by law.’ Coffin v. *140Reichard, 143 F. 2d 443, 445 (CA6 1944).” Id., at 422 (concurring opinion). It follows from this tenet that a restriction on the First Amendment rights of prisoners, like a restriction on the rights of nonprisoners, “can only be justified by a substantial government interest and a showing that the means chosen to effectuate the State's purpose are not unnecessarily restrictive of personal freedoms.” Id., at 423. This does not mean that any expressive conduct that would be constitutionally protected outside a prison is necessarily protected inside; as I also stated in Martinez: “[T]he First Amendment must in each context ‘be applied “in light of the special characteristics of the . . . environment,”' Ready v. James, 408 U. S. 169, 180 (1972), and the exigencies of governing persons in prisons are different from and greater than those in governing persons without.” Id., at 424. But the basic mode of First Amendment analysis — the requirement that restrictions on speech be supported by “reasons imperatively justifying the particular deprivation,” ibid. — should not be altered simply because the First Amendment claimants are incarcerated.
The Court today rejects this analytic framework, at least as it applies to the right of prisoners to associate in something called a prison “union.” 2 In testing restrictions on the exercise of that right the Court asks only whether the restrictions are “rationally related to the . . . objectives of prison administration,” ante, at 129, and whether the reasons offered in defense of the restrictions have been “conclusively shown to be wrong,” ante, at 132. While proclaiming faithfulness to the teaching of Pell v. Procunier, 417 U. S. 817, 822 (1974), that “ ‘a prison inmate retains those First Amendment rights that are not inconsistent with his status as a prisoner,' ” ante, at 125, the Court ultimately upholds the challenged regulations *141on a ground that would apply to any restriction on inmate freedom: they “are consistent with the inmates’ status as prisoners,” ante, at 130.
Nothing in the Court’s opinion justifies its wholesale abandonment of traditional principles of First Amendment analysis. I realize, of course, that “the realities of running a penal institution are complex and difficult,” ante, at 126, and that correctional officers possess considerably more “ ‘professional expertise,’ ” ante, at 128, in prison management than do judges. I do not in any way minimize either the seriousness of the problems or the significance of the expertise. But it does seem to me that “the realities of running” a school or a city are also “complex and difficult,” and that those charged with these tasks — principals, college presidents, mayors, councilmen, and law enforcement personnel — also possess special “professional expertise,” 3 Yet in no First Amendment case of which I am aware has the Court deferred to the judgment of such officials simply because their judgment was “rational.” Cf. Healy v. James, 408 U. S. 169 (1972); Tinker v. Des Moines School Dist., 393 U. S. 503 (1969); Cox v. Louisiana, 379 U. S. 536, 544-551 (1965); Edwards v. South Carolina, 372 U. S. 229 (1963). I do not understand why a different rule should apply simply because prisons are involved.
The reason courts cannot blindly defer to the judgment of prison administrators — or any other officials for that matter— is easily understood. Because the prison administrator’s business is to maintain order, “there inheres the danger that he may well be less responsive than a court — part of an independent branch of government — to the constitutionally protected interests in free expression.” Freedman v. Maryland, 380 U. S. 51, 57-58 (1965). A warden seldom will find himself subject to public criticism or dismissal because he *142needlessly repressed free speech; indeed, neither the public nor the warden will have any way of knowing when repression was unnecessary. But a warden’s job can be jeopardized and public criticism is sure to come should disorder occur. Consequently, prison officials inevitably will err on the side of too little freedom. That this has occurred in the past is made clear by the recent report of the American Bar Association Joint Committee on the Legal Status of Prisoners:
“All organizations including correctional organizations overreact to suggested changes, whether sweeping or merely incremental. . . . [M] any of the fears voiced by prison officials in the 1960s to the growing tide of court determinations invalidating prison regulations have simply not come to pass; indeed, in several instances . . . those groups feared by the prisons in the 1960s have become stablilizing influences in the 1970s.” 4
I do not mean to suggest that the views of correctional officials should be cavalierly disregarded by courts called upon to adjudicate constitutional claims of prisoners. Far from it. The officials’ views “ ‘constitute a body of experience and informed judgment to which courts . . . may properly resort for guidance. The weight of such a judgment in a particular case will depend upon the thoroughness evident in its consideration, the validity of its reasoning . . . and all those factors which give it power to persuade General Electric Co. v. Gilbert, 429 U. S. 125, 142 (1976), quoting Skidmore v. Swift & Co., 323 U. S. 134, 140 (1944). My point is simply that the ultimate responsibility for evaluating the prison officials’ testimony, as well as any other expert testimony, must rest with the courts, which are required to reach an independent judg-*143meat concerning the constitutionality of any restriction on expressive activity.
The approach I advocate is precisely the one this Court has followed in other cases involving the rights of prisoners. In Johnson v. Avery, 393 U. S. 483 (1969), for example, the Court expressly acknowledged the rationality of the rule at issue which prohibited inmate writ writers from aiding fellow prisoners in preparing legal papers, id., at 488. We nevertheless concluded that the rule was unconstitutional because of its impact on prisoners’ right of access to the courts. In Lee v. Washington, 390 U. S. 333 (1968), we did not even inquire whether segregating prisoners by race was rational, although it could be argued that integration in a southern prison would lead to disorder among inmates; we held that in any event segregation was prohibited by the Fourteenth Amendment. And in Bounds v. Smith, 430 U. S. 817 (1977); Wolff v. McDonnell, 418 U. S. 539 (1974); and Cruz v. Beto, 405 U. S. 319 (1972), we followed the approach of Lee. By word and deed, then, we have repeatedly reaffirmed that “a policy of judicial restraint cannot encompass any failure to take cognizance of valid constitutional claims .... When a prison regulation or practice offends a fundamental constitutional guarantee, federal courts will discharge their duty to protect constitutional rights.” Procunier v. Martinez, 416 U. S., at 405.
II
Once it is established that traditional First Amendment principles are applicable in prisoners’-rights cases, the dispute here is easily resolved. The three-judge court not only found that there was “not one scintilla of evidence to suggest that the Union had been utilized to disrupt the operation of the penal institutions,” 409 F. Supp. 937, 944 (EDNC 1976), as the Court acknowledges, ante, at 127 n. 5, it also found no evidence “that the inmates intend to operate [the Union] to *144hamper and interfere with the proper interests of government,” 409 F. Supp., at 944, or that the Union posed a “present danger to security and order,” id., at 945. In the face of these findings, it cannot be argued that the restrictions on the Union are “imperatively justified].”
The regulation barring inmates from soliciting fellow prisoners to join the Union is particularly vulnerable to attack. As the late Judge Craven stated for the court below: “To permit an inmate to join a union and forbid his inviting others to join borders on the irrational.” Id., at 943. The irrationality of the regulation is perhaps best demonstrated by the fact that the Court does not defend it; rather, as my Brother Stevens suggests, ante, at 138-139, the Court defends some hypothetical regulation banning “ 'an invitation to collectively engage in a legitimately prohibited activity.’ Ante, at 132”; see also ante, at 129 (discussing ban on “concerted group activity, or solicitation therefor”). Because the actual regulation at issue here needlessly bars solicitation for an activity — -joining the Union — which is not and presumably could not be prohibited,5 I would hold it unconstitutional.
Once the rule outlawing solicitation is invalidated, the prohibition on bulk mailing by the Union must fall with it. Since North Carolina allows the Union to mail its newsletters to prisoners individually, the State cannot claim that the bulk mail rule serves to keep “subversive material” out of the prison. Rather, the primary purpose of the rule must be to supplement the ban on solicitation;6 overturning that ban *145would sap all force from the rationale for excluding bulk mailings. The exclusion would then be left as one that unnecessarily increases the cost to the Union of exercising its First Amendment rights7 while allowing other inmate groups such as the Jaycees to exercise their rights at a lower price. It would, therefore, be plainly unconstitutional.
The regulation prohibiting the Union from holding meetings within the prison is somewhat more justifiable than the regulations previously considered. Once the Union is permitted to hold meetings it will become operational within the prisons. Appellants’ fears that the leaders of an operating union “would be in a position to misuse their influence” and that the Union itself could engage in disruptive, concerted activities or increase tension within the prisons, App. 121, are not entirely fanciful. It is important to note, however, that appellee’s two expert witnesses, both correctional officers who had dealt with inmate reform organizations, testified that such groups actually play a constructive role in their prisons, id., at 38, 90-95. The weight of professional opinion seems to favor recognizing such groups.8 Moreover, the risks appellants fear are inherent in any inmate organization, no matter how innocuous its stated goals; indeed, even without any organizations some inmates inevitably will become leaders capable *146of “misus[ing] their influence,” id., at 84-86, 102-103,9 and some concerted activity can still occur, id., at 118-119. .
But even if the risks posed by the Union were unique to- it, and even if appellants’ fear of the Union were more widely shared by other professionals, the prohibition on Union meetings still could not survive constitutional attack. The central lesson of over a half century of First Amendment adjudication is that freedom is sometimes a hazardous enterprise, and that the Constitution requires the State to bear certain risks to preserve our liberty. See, e. g., Whitney v. California, 274 U. S. 357, 375-378 (1927) (Brandéis, J., concurring); Terminiello v. Chicago, 337 U. S. 1 (1949); Tinker v. Des Moines School Dist., 393 U. S. 503 (1969). As the ABA Joint Committee, supra, put it: “The doubts and risks raised by creating a humane and open prison must* be accepted as a cost of our society; democracy is self-definitionally a risk-taking form of government.” 10 To my mind, therefore, the fact that appellants have not acted wholly irrationally in banning Union meetings is not dispositive. Rather, I believe that where, as here, meetings would not pose an immediate and substantial threat to the security or rehabilitative functions of the prisons, the First Amendment guarantees Union members the right to associate freely, and the Fourteenth Amendment guarantees them the right to be treated as favorably as members of other inmate organizations. The State can surely regulate the time, place, and manner of the meetings, and perhaps can monitor them to assure that disruptions are not planned, but the State cannot outlaw such assemblies altogether.
*147III
If the mode of analysis adopted in today’s decision were to be generally followed, prisoners eventually would be stripped of all constitutional rights, and would retain only those privileges that prison officials, in their “informed discretion,” deigned to recognize. The sole constitutional constraint on prison officials would be a requirement that they act rationally. Ironically, prisoners would be left with a right of access to the courts, see Bounds v. Smith, 430 U. S. 817 (1977); Johnson v. Avery, 393 U. S. 483 (1969), but no substantive rights to assert once they get there. I cannot believe that the Court that decided Bounds and Johnson — the Court that has stated that “[tjhere is no iron curtain drawn between the Constitution and the prisons of this country,” Wolff v. McDonnell, 418 U. S., at 555-556, and that “[a] prison inmate retains those First Amendment rights that are not inconsistent with his status as a prisoner,” Pell v. Procunier, 417 U. S., at 822 — intends to allow this to happen. I therefore believe that the tension between today’s decision and our prior cases ultimately will be resolved, not by the demise of the earlier cases, but by the recognition that the decision today is an aberration, a manifestation of the extent to which the very phrase “prisoner union” is threatening to those holding traditional conceptions of the nature of penal institutions.
I respectfully dissent.
For brief exposition of the “hands-off” doctrine and its demise, see Fox, The First Amendment Rights of Prisoners, 63 J. Crim. L. C. & P. S. 162 (1972).
That the First Amendment protects the right to associate is by now well established. See, e. g., Kusper v. Pontikes, 414 U. S. 51 (1973); NAACP v. Alabama, 357 U. S. 449 (1958).
Similarly, prison administrators, principals, college presidents, and the like “must be permitted to act before the time when they can compile a dossier on the eve of a riot,” Ante, at 132-133.
ABA Joint Committee on the Legal Status of Prisoners, The Legal Status of Prisoners (Tent. Draft 1977), in 14 Am. Crim. L. Rev. 377, 419 (1977) (hereafter ABA Joint Committee report).
1 express no view concerning the extent to which orderly, concerted activities are protected in prison. This issue has been addressed at length by the ABA Joint Committee report, Standard § 6.4 and Commentary.
The only other justification offered for the rule is to prevent contraband from being smuggled into the prisons. Nothing in the record remotely suggests that the outside personnel associated with the Union would use bulk mailing for this purpose. Moreover, the solution to the *145alleged contraband danger is to inspect the bulk mailings, not to prohibit them.
Contrary to the Court’s assertion, ante, at 130-131, free speech values most definitely are implicated by a regulation whose purpose and effect is to make the exercise of First Amendment rights costly. Cf., e. g., Murdock v. Pennsylvania, 319 U. S. 105 (1943); Grosjean v. American Press Co., 297 U. S. 233 (1936).
See ABA Joint Committee report, Standard §6.4 and Commentary; S. Krantz, R. Bell, J. Brant, & M. Magruder, Model Rules and Regulations on Prisoners’ Rights and Responsibilities, Rules IA-lb, IA-5 and Commentary (1973); National Advisory Commission on Criminal Justice Standards and Goals, Corrections, Standard 2.15 and Commentary, pp. 58-61 (1973).
See also Note, Bargaining in Correctional Institutions: Restructuring the Relation between the Inmate and the Prison Authority, 81 Yale L. J. 726 (1972). The concern over inmate leadership has been advanced to oppose numerous prison reforms. E. g., Johnson v. Avery, 393 U. S. 483, 499 (1969) (White, J., dissenting); Saxbe v. Washington Post Co., 417 U. S. 843, 866-869 (1974) (Powell, J., dissenting) (rejecting argument).
ABA Joint Committee report 419.