delivered the opinion of the Court.
In Bounds v. Smith, 430 U. S. 817 (1977), we held that “the fundamental constitutional right of access to the courts requires prison authorities to assist inmates in the preparation and filing of meaningful legal papers by providing prisoners with adequate law libraries or adequate assistance from persons trained in the law.” Id., at 828. Petitioners, who are officials of the Arizona Department of Corrections (ADOC), contend that the United States District Court for the District of Arizona erred in finding them in violation of Bounds, and that the court’s remedial order exceeded lawful authority.
I
Respondents are 22 inmates of various prisons operated by ADOC. In January 1990, they filed this class action “on behalf of all adult prisoners who are or will be incarcerated by the State of Arizona Department of Corrections,” App. 22, alleging that petitioners were “depriving [respondents] of their rights of access to the courts and counsel protected by the First, Sixth, and Fourteenth Amendments,” id., at 34. Following a 3-month bench trial, the District Court ruled in favor of respondents, finding that “[prisoners have a constitutional right of access to the courts that is adequate, effective and' meaningful,” 834 F. Supp. 1553, 1566 (1992), citing Bounds, supra, at 822, and that “[ADOC’s] system fails to comply with constitutional standards,” 834 F. Supp., at 1569. The court identified a variety of shortcomings of the ADOC system, in matters ranging from the training of library staff, to the updating of legal materials, to the availability of photocopying services. In addition to these gen*347eral findings, the court found that two groups of inmates were particularly affected by the system’s inadequacies: “[l]ockdown prisoners” (inmates segregated from the general prison population for disciplinary or security reasons), who “are routinely denied physical access to the law library” and “experience severe interference with their access to the courts,” id., at 1556; and illiterate or non-English-speaking inmates, who do not receive adequate legal assistance, id., at 1558.
Having thus found liability, the court appointed a Special Master “to investigate and report about” the appropriate relief — that is (in the court’s view), “how best to accomplish the goal of constitutionally adequate inmate access to the courts.” App. to Pet. for Cert. 87a. Following eight months of investigation, and some degree of consultation with both parties, the Special Master lodged with the court a proposed permanent injunction, which the court proceeded to adopt, substantially unchanged. The 25-page injunctive order, see id., at 61a-85a, mandated sweeping changes designed to ensure that ADOC would “provide meaningful access to the Courts for all present and future prisoners,” id., at 61a. It specified in minute detail the times that libraries were to be kept open, the number of hours of library use to which each inmate was entitled (10 per week), the minimal educational requirements for prison librarians (a library science degree, law degree, or paralegal degree), the content of a videotaped legal-research course for inmates (to be prepared by persons appointed by the Special Master but funded by ADOC), and similar matters. Id., at 61a, 67a, 71a. The injunction addressed the court’s concern for lockdown prisoners by ordering that “ADOC prisoners in all housing areas and custody levels shall be provided regular and comparable visits to the law library,” except that such visits “may be postponed on an individual basis because of the prisoner’s documented inability to use the law library without creating *348a threat to safety or security, or a physical condition if determined by medical personnel to prevent library use.” Id., at 61a. With respect to illiterate and non-English-speaking inmates, the injunction declared that they were entitled to “direct assistance” from lawyers, paralegals, or “a sufficient number of at least minimally trained prisoner Legal Assistants”; it enjoined ADOC that “[particular steps must be taken to locate and train bilingual prisoners to be Legal Assistants.” Id., at 69a-70a.
Petitioners sought review in the Court of Appeals for the Ninth Circuit, which refused to grant a stay prior to argument. We then stayed the injunction pending filing and disposition of a petition for a writ of certiorari. 511 U. S. 1066 (1994). Several months later, the Ninth Circuit affirmed both the finding of a Bounds violation and, with minor exceptions not important here, the terms of the injunction. 48 F. 3d 1261 (1994). We granted certiorari, 514 U. S. 1126 (1995).
II
Although petitioners present only one question for review, namely, whether the District Court’s order “exceeds the constitutional requirements set forth in Bounds,” Brief for Petitioners (i), they raise several distinct challenges, including renewed attacks on the court’s findings of Bounds violations with respect to illiterate, non-English-speaking, and lock-down prisoners, and on the breadth of the injunction. But their most fundamental contention is that the District Court’s findings of injury were inadequate to justify the finding of systemwide injury and hence the granting of sys-temwide relief. This argument has two related components. First, petitioners claim that in order to establish a violation of Bounds, an inmate must show that the alleged inadequacies of a prison’s library facilities or legal assistance program caused him “actual injury” — that is, “actual prejudice with respect to contemplated or existing litigation, such as the inability to meet a filing deadline or to present a claim.” *349Brief for Petitioners 30.1 Second, they claim that the District Court did not find enough instances of actual injury to warrant systemwide relief. We agree that the success of respondents’ systemic challenge was dependent on their ability to show widespread actual injury, and that the court’s failure to identify anything more than isolated instances of actual injury renders its finding of a systemic Bounds violation invalid.
A
The requirement that an inmate alleging a violation of Bounds must show actual injury derives ultimately from the doctrine of standing, a constitutional principle that prevents courts of law from undertaking tasks assigned to the political branches. See Allen v. Wright, 468 U. S. 737, 750-752 (1984); Valley Forge Christian College v. Americans United for Separation of Church and State, Inc., 454 U. S. 464, 471-476 (1982). It is the role of courts to provide relief to claimants, in individual or class actions, who have suffered, or will imminently suffer, actual harm; it is not the role of courts, but that of the political branches, to shape the institutions of government in such fashion as to comply with the laws and the Constitution. In the context of the present case: It is for the courts to remedy past or imminent official interference with individual inmates’ presentation of claims to the courts; it is for the political branches of the State and Federal Governments to manage prisons in such fashion that official interference with the presentation of claims will not occur. *350Of course, the two roles briefly and partially coincide when a court, in granting relief against actual harm that has been suffered, or that will imminently be suffered, by a particular individual or class of individuals, orders the alteration of an institutional organization or procedure that causes the harm. But the distinction between the two roles would be obliterated if, to invoke intervention of the courts, no actual or imminent harm were needed, but merely the status of being subject to a governmental institution that was not organized or managed properly. If — to take another example from prison life — a healthy inmate who had suffered no deprivation of needed medical treatment were able to claim violation of his constitutional right to medical care, see Estelle v. Gamble, 429 U. S. 97, 103 (1976), simply on the ground that the prison medical facilities were inadequate, the essential distinction between judge and executive would have disappeared: it would have become the function of the courts to assure adequate medical care in prisons.
The foregoing analysis would not be pertinent here if, as respondents seem to assume, the right at issue — the right to which the actual or threatened harm must pertain — were the right to a law library or to legal assistance. But Bounds established no such right, any more than Estelle established a right to a prison hospital. The right that Bounds acknowledged was the (already well-established) right of access to the courts. E. g., Bounds, 430 U. S., at 817, 821, 828. In the cases to which Bounds traced its roots, we had protected that right by prohibiting state prison officials from actively interfering with inmates’ attempts to prepare legal documents, e. g., Johnson v. Avery, 393 U. S. 483, 484, 489-490 (1969), or file them, e. g., Ex parte Hull, 312 U. S. 546, 547-549 (1941), and by requiring state courts to waive filing fees, e. g., Burns v. Ohio, 360 U. S. 252, 258 (1959), or transcript fees, e. g., Griffin v. Illinois, 351 U. S. 12, 19 (1956), for indigent inmates. Bounds focused on the same entitlement of access to the courts. Although it affirmed a court order *351requiring North Carolina to make law library facilities available to inmates, it stressed that that was merely “one constitutionally acceptable method to assure meaningful access to the courts,” and that “our decision here ... does not foreclose alternative means to achieve that goal.” 430 U. S., at 830. In other words, prison law libraries and legal assistance programs are not ends in themselves, but only the means for ensuring “a reasonably adequate opportunity to present claimed violations of fundamental constitutional rights to the courts.” Id., at 825.
Because Bounds did not create an abstract, freestanding right to a law library or legal assistance, an inmate cannot establish relevant actual injury simply by establishing that his prison’s law library or legal assistance program is subpar in some theoretical sense. That would be the precise analog of the healthy inmate claiming constitutional violation because of the inadequacy of the prison infirmary. Insofar as the right vindicated by Bounds is concerned, “meaningful access to the courts is the touchstone,” id., at 823 (internal quotation marks omitted), and the inmate therefore must go one step further and demonstrate that the alleged shortcomings in the library or legal assistance program hindered his efforts to pursue a legal claim. He might show, for example, that a complaint he prepared was dismissed for failure to satisfy some technical requirement which, because of deficiencies in the prison’s legal assistance facilities, he could not have known. Or that he had suffered arguably actionable harm that he wished to bring before the courts, but was so stymied by inadequacies of the law library that he was unable even to file a complaint.
Although Bounds itself made no mention of an actual-injury requirement, it can hardly be thought to have eliminated that constitutional prerequisite. And actual injury is apparent on the face of almost all the opinions in the 35-year line of access-to-courts cases on which Bounds relied, see id., *352at 821-825.2 Moreover, the assumption of an actual-injury requirement seems to us implicit in the opinion’s statement that “we encourage local experimentation” in various methods of assuring access to the courts. Id., at 832. One such experiment, for example, might replace libraries with some minimal access to legal advice and a system of court-provided forms such as those that contained the original complaints in two of the more significant inmate-initiated cases in recent years, Sandin v. Conner, 515 U. S. 472 (1995), and Hudson v. McMillian, 503 U. S. 1 (1992) — forms that asked the inmates to provide only the facts and not to attempt any legal analysis. We hardly think that what we meant by “experimenting” with such an alternative was simply announcing it, whereupon suit would immediately lie to declare it theoretically inadequate and bring the experiment to a close. We think we envisioned, instead, that the new *353program would remain in place at least until some inmate could demonstrate that a nonfrivolous3 legal claim had been frustrated or was being impeded.4
*354It must be acknowledged that several statements in Bounds went beyond the right of access recognized in the earlier cases on which it relied, which was a right to bring to court a grievance that the inmate wished to present, see, e. g., Ex parte Hull, 312 U. S., at 547-548; Griffin v. Illinois, 351 U. S., at 13-16; Johnson v. Avery, 393 U. S., at 489. These statements appear to suggest that the State must enable the prisoner to discover grievances, and to litigate effectively once in court. See Bounds, 430 U. S., at 825-826, and n. 14. These elaborations upon the right of access to the courts have no antecedent in our pre-Bounds cases, and we now disclaim them. To demand the conferral of such sophisticated legal capabilities upon a mostly uneducated and indeed largely illiterate prison population is effectively to demand permanent provision of counsel, which we do not believe the Constitution requires.
Finally, we must observe that the injury requirement is not satisfied by just any type of frustrated legal claim. Nearly all of the access-to-courts cases in the Bounds line involved attempts by inmates to pursue direct appeals from the convictions for which they were incarcerated, see Douglas v. California, 372 U. S. 353, 354 (1963); Burns v. Ohio, 360 U. S., at 253, 258; Griffin v. Illinois, supra, at 13, 18; Cochran v. Kansas, 316 U. S. 255, 256 (1942), or habeas petitions, see Johnson v. Avery, supra, at 489; Smith v. Bennett, 365 U. S. 708, 709-710 (1961); Ex parte Hull, supra, at 547-548. In Wolff v. McDonnell, 418 U. S. 539 (1974), we extended this universe of relevant claims only slightly, to “civil rights actions” — i. e., actions under 42 U. S. C. § 1983 to vindicate “basic constitutional rights.” 418 U. S., at 579. Significantly, we felt compelled to justify even this slight extension of the right of access to the courts, stressing that “the demarcation line between civil rights actions and ha-*355beas petitions is not always clear,” and that “[i]t is futile to contend that the Civil Rights Act of 1871 has less importance in our constitutional scheme than does the Great Writ.” Ibid. The prison law library imposed in Bounds itself was far from an all-subject facility. In rejecting the contention that the State’s proposed collection was inadequate, the District Court there said:
“This Court does not feel inmates need the entire U. S. Code Annotated. Most of that code deals with federal laws and regulations that would never involve a state prisoner.. . .
“It is also the opinion of this Court that the cost of N. C. Digest and Modern Federal Practice Digest will surpass the usefulness of these research aids. They cover mostly areas not of concern to inmates.”5 Supplemental App. to Pet. for Cert. in Bounds v. Smith, O. T. 1976, No. 75-915, p. 18.
In other words, Bounds does not guarantee inmates the wherewithal to transform themselves into litigating engines capable of filing everything from shareholder derivative actions to slip-and-fall claims. The tools it requires to be provided are those that the inmates need in order to attack their sentences, directly or collaterally, and in order to challenge the conditions of their confinement. Impairment of any other litigating capacity is simply one of the incidental (and perfectly constitutional) consequences of conviction and incarceration.
*356B
Here the District Court identified only two instances of actual injury. In describing ADOC’s failures with respect to illiterate and non-English-speaking prisoners, it found that “[a]s a result of the inability to receive adequate legal assistance, prisoners who are slow readers have had their cases dismissed with prejudice,” and that “[ojther prisoners have been unable to file legal actions.” 834 F. Supp., at 1558. Although the use of the plural suggests that several prisoners sustained these actual harms, the court identified only one prisoner in each instance. Id., at 1558, nn. 37 (lawsuit of inmate Bartholic dismissed with prejudice), 38 (inmate Harris unable to file a legal action).
Petitioners contend that “any lack of access experienced by these two inmates is not attributable to unconstitutional State policies,” because ADOC “has met its constitutional obligations.” Brief for Petitioners 32, n. 22. The claim appears to be that all inmates, including the illiterate and non-English speaking, have a right to nothing more than “physical access to excellent libraries, plus help from legal assistants and law clerks.” Id., at 35. This misreads Bounds, which as we have said guarantees no particular methodology but rather the conferral of a capability — the capability of bringing contemplated challenges to sentences or conditions of confinement before the courts. When any inmate, even an illiterate or non-English-speaking inmate, shows that an actionable claim of this nature which he desired to bring has been lost or rejected, or that the presentation of such a claim is currently being prevented, because this capability of filing suit has not been provided, he demonstrates that the State has failed to furnish “adequate law libraries or adequate assistance from persons trained in the law,” Bounds, 430 U. S., at 828 (emphasis added). Of course, we leave it to prison officials to determine how best to ensure that inmates with language problems have a reasonably adequate opportunity to file nonfrivolous legal claims challenging their convictions or conditions of confinement. But it is *357that capability, rather than the capability of turning pages in a law library, that is the touchstone.
C
Having rejected petitioners’ argument that the injuries suffered by Bartholic and Harris do not count, we turn to the question whether those injuries, and the other findings of the District Court, support the injunction ordered in this case. The actual-injury requirement would hardly serve the purpose we have described above — of preventing courts from undertaking tasks assigned to the political branches— if once a plaintiff demonstrated harm from one particular inadequacy in government administration, the court were authorized to remedy all inadequacies in that administration. The remedy must of course be limited to the inadequacy that produced the injury in fact that the plaintiff has established. See Missouri v. Jenkins, 515 U. S. 70, 88, 89 (1995) (“[T]he nature of the . . . remedy is to be determined by the nature and scope of the constitutional violation” (citation and internal quotation marks omitted)).
This is no less true with respect to class actions than with respect to other suits. “That a suit may be a class action ... adds nothing to the question of standing, for even named plaintiffs who represent a class 'must allege and show that they personally have been injured, not that injury has been suffered by other, unidentified members of the class to which they belong and which they purport to represent.’ ” Simon v. Eastern Ky. Welfare Rights Organization, 426 U. S. 26, 40, n. 20 (1976), quoting Warth v. Seldin, 422 U. S. 490, 502 (1975). The general allegations of the complaint in the present case may well have sufficed to claim injury by named plaintiffs, and hence standing to demand remediation, with respect to various alleged inadequacies in the prison system, including failure to provide adequate legal assistance to non-English-speaking inmates and lockdown prisoners. That point is irrelevant now, however, for we are beyond the pleading stage.
*358“Since they are not mere pleading requirements, but rather an- indispensable part of the plaintiff’s case, each element [of standing] must be supported in the same way as any other matter on which the plaintiff bears the burden of proof, i. e., with the manner and degree of evidence required at the successive stages of the litigation. At the pleading stage, general factual allegations of injury resulting from the defendant’s conduct may suffice, for on a motion to dismiss we presume that general allegations embrace those specific facts that are necessary to support the claim. In response to a summary judgment motion, however, the plaintiff can no longer rest on such mere allegations, but must set forth by affidavit or other evidence specific facts, which for purposes of the summary judgment motion will be taken to be true. And at the final stage, those facts (if controverted) must be supported adequately by the evidence adduced at trial.” Lujan v. Defenders of Wildlife, 504 U. S. 555, 561 (1992) (citations and internal quotation marks omitted).
After the trial in this case, the court found actual injury on the part of only one named plaintiff, Bartholic; and the cause of that injury — the inadequacy which the suit empowered the court to remedy — was failure of the prison to provide the special services that Bartholic would have needed, in light of his illiteracy, to avoid dismissal of his case. At the outset, therefore, we can eliminate from the proper scope of this injunction provisions directed at special services or special facilities required by non-English speakers, by prisoners in lockdown, and by the inmate population at large. If inadequacies of this character exist, they have not been found to have harmed any plaintiff in this lawsuit, and hence were not the proper object of this District Court’s remediation.6
*359As to remediation of the inadequacy that caused Bartholic’s injury, a further question remains: Was that inadequacy widespread enough to justify systemwide relief? The only findings supporting the proposition that, in all of ADOC’s facilities, an illiterate inmate wishing to file a claim would be unable to receive the assistance necessary to do so were (1) the finding with respect to Bartholic, at the Florence facility, and (2) the finding that Harris, while incarcerated at Perryville, had once been “unable to file [a] legal actio[n].” 834 F. Supp., at 1558. These two instances were a patently inadequate basis for a conclusion of systemwide violation and imposition of systemwide relief. See Dayton Bd. of Ed. v. Brinkman, 433 U. S. 406, 417 (1977) (“[I]nstead of tailoring a remedy commensurate with the three specific violations, the Court of Appeals imposed a systemwide remedy going beyond their scope”); id., at 420 (“[O]nly if there has been a systemwide impact may there be a systemwide remedy”); *360Califano v. Yamasaki, 442 U. S. 682, 702 (1979) (“[T]he scope of injunctive relief is dictated by the extent of the violation established, not by the geographical extent of the plaintiff class”).
To be sure, the District Court also noted that “the trial testimony . . . indicated that there are prisoners who are unable to research the law because of their functional illiteracy,” 834 F. Supp., at 1558. As we have discussed, however, the Constitution does not require that prisoners (literate or illiterate) be able to conduct generalized research, but only that they be able to present their grievances to the courts— a more limited capability that can be produced by a much more limited degree of legal assistance. Apart from the dismissal of Bartholic’s claim with prejudice, and Harris’s inability to file his claim, there is no finding, and as far as we can discern from the record no evidence, that in Arizona prisons illiterate prisoners cannot obtain the minimal help necessary to file particular claims that they wish to bring before the courts. The constitutional violation has not been shown to be systemwide, and granting a remedy beyond what was necessary to provide relief to Harris and Bartholic was therefore improper.7
*361III
There are further reasons why the order here cannot stand. We held in Turner v. Safley, 482 U. S. 78 (1987), that a prison regulation impinging on inmates’ constitutional rights “is valid if it is reasonably related to legitimate peno-logical interests.” Id., at 89. Such a deferential standard is necessary, we explained,
“if 'prison administrators . . . , and not the courts, [are] to make the difficult judgments concerning institutional operations.’ Subjecting the day-to-day judgments of prison officials to an inflexible strict scrutiny analysis would seriously hamper their ability to anticipate security problems and to adopt innovative solutions to the intractable problems of prison administration.” Ibid. (citation omitted), quoting Jones v. North Carolina Prisoners’ Labor Union, Inc., 433 U. S. 119, 128 (1977).
These are the same concerns that led us to encourage “local experimentation” in Bounds, see supra, at 352, and we think it quite obvious that Bounds and Turner must be read in pari materia.
The District Court here failed to accord adequate deference to the judgment of the prison authorities in at least three significant respects. First, the court concluded that ADOC’s restrictions on lockdown prisoners’ access to law libraries were unjustified. Turner’s principle of deference has special force with regard to that issue, since the inmates in lockdown include “the most dangerous and violent prisoners in the Arizona prison system,” and other inmates presenting special disciplinary and security concerns. Brief for Petitioners 5. The District Court made much of the fact *362that lockdown prisoners routinely experience delays in receiving legal materials or legal assistance, some as long as 16 days, 834 F. Supp., at 1557, and n. 23, but so long as they are the product of prison regulations reasonably related to legitimate penological interests, such delays are not of constitutional significance, even where they result in actual injury (which, of course, the District Court did not find here).
Second, the injunction imposed by the District Court was inordinately — indeed, wildly — intrusive. There is no need to belabor this point. One need only read the order, see App. to Pet. for Cert. 61a-85a, to appreciate that it is the ne plus ultra of what our opinions have lamented as a court’s “in the name of the Constitution, becoming] . . . enmeshed in the minutiae of prison operations.” Bell v. Wolfish, 441 U. S. 520, 562 (1979).
Finally, the order was developed through a process that failed to give adequate consideration to the views of state prison authorities. We have said that “[t]he strong considerations of comity that require giving a state court system that has convicted a defendant the first opportunity to correct its own errors ... also require giving the States the first opportunity to correct the errors made in the internal administration of their prisons.” Preiser v. Rodriguez, 411 U. S. 475, 492 (1973). For an illustration of the proper procedure in a case such as this, we need look no further than Bounds itself. There, after granting summary judgment for the inmates, the District Court refrained from “ ‘dictating] precisely what course the State should follow.’” Bounds, 430 U. S., at 818. Rather, recognizing that “determining the ‘appropriate relief to be ordered ... presents a difficult problem,’ ” the court “ ‘charge[d] the Department of Correction with the task of devising a Constitutionally sound program’ to assure inmate access to the courts.” Id., at 818-819. The State responded with a proposal, which the District Court ultimately approved with minor changes, after considering ob*363jections raised by the inmates. Id., at 819-820. We praised this procedure, observing that the court had “scrupulously respected the limits on [its] role,” by “not . . . thrust[ing] itself into prison administration” and instead permitting “[p]rison administrators [to] exercis[e] wide discretion within the bounds of constitutional requirements.” Id., at 832-833.
As Bounds was an exemplar of what should be done, this case is a model of what should not. The District Court totally failed to heed the admonition of Preiser. Having found a violation of the right of access to the courts, it conferred upon its special master, a law professor from Flushing, New York, rather than upon ADOC officials, the responsibility for devising a remedial plan. To make matters worse, it severely limited the remedies that the master could choose. Because, in the court’s view, its order in an earlier access-to-courts case (an order that adopted the recommendations of the same special master) had “resolved successfully” most of the issues involved in this litigation, the court instructed that as to those issues it would implement the earlier order statewide, “with any modifications that the parties and Special Master determine are necessary due to the particular circumstances of the prison facility.” App. to Pet. for Cert. 88a (footnote omitted). This will not do. The State was entitled to far more than an opportunity for rebuttal, and on that ground alone this order would have to be set aside.8
*364* * *
For the foregoing reasons, we reverse the judgment of the Court of Appeals and remand the case for further proceedings consistent with this opinion.
It is so ordered.
Respondents contend that petitioners failed properly to present their “actual injury” argument to the Court of Appeals. Brief for Respondents 25-26. Our review of petitioners’ briefs before that court leads us to conclude otherwise, and in any event, as we shall discuss, the point relates to standing, which is jurisdictional and not subject to waiver. See United States v. Hays, 515 U. S. 737, 742 (1995); FW/PBS, Inc. v. Dallas, 493 U. S. 215, 230-231 (1990). Justice Souter recognizes the jurisdictional nature of this point, post, at 394, which is difficult to reconcile with his view that we should not “reach out to address” it, ibid.
Justice Stevens suggests that Ex parte Hull, 312 U. S. 546 (1941), establishes that even a lost frivolous claim establishes standing to complain of a denial of access to courts, see post, at 408-409. As an initial matter, that is quite impossible, since standing was neither challenged nor discussed in that case, and we have repeatedly held that the existence of unaddressed jurisdictional defects has no precedential effect. See, e. g., Federal Election Comm’n v. NRA Political Victory Fund, 513 U. S. 88, 97 (1994); United States v. More, 3 Cranch 159, 172 (1805) (Marshall, C. J.) (statement at oral argument). On the merits, however, it is simply not true that the prisoner’s claim in Hull was frivolous. We rejected it because it had been procedurally defaulted by, inter alia, failure to object at trial and failure to include a transcript with the petition, 312 U. S., at 551. If all procedurally defaulted claims were frivolous, Kule 11 business would be brisk indeed. Justice Stevens’s assertion that “we held that the smuggled petition had insufficient merit even to require an answer from the State,” post, at 408-409, is misleading. The attorney general of Michigan appeared in the ease, and our opinion discussed the merits of the claim at some length, see 312 U. S., at 549-551. The posture of the case was such, however, that we treated the claim “as a motion for leave to file a petition for writ of habeas corpus,” id., at 550; after analyzing petitioner’s case, we found it “insufficient to compel an order requiring the warden to answer,” id., at 551 (emphasis added). That is not remotely equivalent to finding that the underlying claim was frivolous.
Justice Souter believes that Bounds v. Smith, 430 U. S. 817 (1977), guarantees prison inmates the right to present frivolous claims — the determination of which suffices to confer standing, he says, because it assumes that the dispute “ ‘will be presented in an adversary context and in a form historically viewed as capable o.f judicial resolution,’” post, at 398-399, quoting Flast v. Cohen, 392 U. S. $3,101 (1968). This would perhaps have seemed like good law at the time of Flast, but our later opinions have made it explicitly clear that Flast erred in assuming that assurance of “serious and adversarial treatment” was the only value protected by standing. See, e. g., United. States v. Richardson, 418 U. S. 166, 176-180 (1974); Schlesinger v. Reservists Comm, to Stop the War, 418 U. S. 208, 220-223 (1974). Flast failed to recognize that this doctrine has a separation-of-powers component, which keeps courts within certain traditional bounds vis-a-vis the other branches, concrete adverseness or not. That is where the “actual injury” requirement comes from. Not everyone who can point to some “concrete” act and is “adverse” can call in the courts to examine the propriety of executive action, but only someone who has been actually injured. Depriving someone of an arguable (though not yet established) claim inflicts actual injury because it deprives him of something of value — arguable claims are settled, bought, and sold. Depriving someone of a frivolous claim, on the other hand, deprives him of nothing at all, except perhaps the punishment of Federal Rule of Civil Procedure 11 sanctions.
Justice Souter suggests that he would waive this actual-injury requirement in cases “involving substantial, systemic deprivation of access to court” — that is, in cases involving “ ‘a direct, substantial and continuous ... limit on legal materials,’ ” “total denial of access to a library,” or “ ‘[a]n absolute deprivation of access to all legal materials,’ ” post, at 401, and 400, n. 2. That view rests upon the expansive understanding of Bounds that we have repudiated. Unless prisoners have a freestanding right to libraries, a showing of the sort Justice Souter describes would establish no relevant injury in fact, i. e., injury-in-fact caused by the violation of legal right. See Allen v. Wright, 468 U. S. 737, 751 (1984). Denial of access to the courts could not possibly cause the harm of inadequate libraries, but only the harm of lost, rejected, or impeded legal claims.
Of course, Justice Souter’s proposed exception is unlikely to be of much real-world significance in any event. Where the situation is so extreme as to constitute “an absolute deprivation of access to all legal *354materials,” finding a prisoner with a claim affected by this extremity will probably be easier than proving the extremity.
The District Court order in this case, by contrast, required ADOC to stock each library with, inter alia, the Arizona Digest, the Modern Federal Practice Digest, Corpus Juris Secundum, and a full set of the United States Code Annotated, and to provide a 30-40 hour videotaped legal research course covering “relevant tort and civil law, including immigration and family issues.” App. to Pet. for Cert. 69a, 71a; 834 F. Supp. 1553, 1561-1562 (Ariz. 1992).
Justice Stevens concludes, in gross, that Bartholic’s and Harris’s injuries are “sufficient to satisfy any constitutional [standing] concerns,” post, at 408. But standing is not dispensed in gross. If the right to complain *359of one administrative deficiency automatically conferred the right to complain of all administrative deficiencies, any citizen aggrieved in one respect could bring the whole structure of state administration before the courts for review. That is of course not the law. As we have said, “[n]or does a plaintiff who has been subject to injurious conduct of one kind possess by virtue of that injury the necessary stake in litigating conduct of another kind, although similar, to which he has not been subject.” Blum v. Yaretsky, 457 U. S. 991, 999 (1982). As even Justice Souter concedes, the inability of respondents to produce any evidence of actual injury to other than illiterate inmates (Bartholic and Harris) “disposed] of the challenge to remedial orders insofar as they touch non-English speakers and lock-down prisoners.” Post, at 395.
Contrary to Justice Stevens’s suggestion, see post, at 408, n. 4, our holding that respondents lacked standing to complain of injuries to non-English speakers and lockdown prisoners does not amount to “a conclusion that the class was improper.” The standing determination is quite separate from certification of the class. Again, Blum proves the point: In that case, we held that a class of “ ‘all residents of skilled nursing and health related nursing facilities in New York State who are recipients of Medicaid benefits’ ” lacked standing to challenge transfers to higher levels of care, even though they had standing to challenge discharges and transfers to lower levels; but we did not disturb the class definition. See 457 U. S., at 997, n. 11, 999-1002.
Our holding regarding the inappropriateness of systemwide relief for illiterate inmates does not rest upon the application of standing rules, but rather, like Justice Souter’s conclusion, upon “the respondents’ failure to prove that denials of access to illiterate prisoners pervaded the State’s prison system,” post, at 397. In one respect, however, Justice Souter’s view of this issue differs from ours. He believes that systemwide relief would have been appropriate “[h]ad the findings shown libraries in shambles throughout the prison system,” ibid. That is consistent with his view, which we have rejected, that lack of access to adequate library facilities qualifies as relevant injury in fact, see n. 4, supra.
Contrary to Justice Souter’s assertion, post, at 397, the issue of sys-temwide relief has nothing to do with the law governing class actions. Whether or not a class of plaintiffs with frustrated nonfrivolous claims exists, and no matter how extensive this class may be, unless it was established that violations with respect to that class occurred in all institutions of Arizona’s system, there was no basis for a remedial decree imposed *361upon all those institutions. However inadequate the library facilities may be as a theoretical matter, various prisons may have other means (active assistance from “jailhouse lawyers,” complaint forms, etc.) that suffice to prevent the legal harm of denial of access to the courts. Courts have no power to presume and remediate harm that has not been established.
Justice Stevens believes that the State of Arizona “is most to blame for the objectionable character of the final [injunctive] order,” post, at 411, for two reasons: First, because of its lack of cooperation in prison litigation three to five years earlier before the same judge, see Gluth v. Kangas, 773 F. Supp. 1309 (Ariz. 1988). But the rule that federal courts must “giv[e] the States the first opportunity to-correct the errors made in the internal administration of their prisons,” Preiser v. Rodriguez, 411 U. S. 475, 492 (1973), is not to be set aside when a judge decides that a State was insufficiently cooperative in a different, earlier case. There was no indication of obstructive tactics by the State in the present case, from which one ought to have concluded that the State had learned its lesson. Second, *364Justice Stevens contends that the State failed vigorously to oppose application of the Gluth methodology to the present litigation. But surely there was no reasonable doubt that the State objected to that methodology. Justice Stevens demands from the State, we think, an unattainable degree of courage and foolishness in insisting that, having been punished for its recalcitrance in the earlier case by the imposition of the Gluth methodology, it antagonize the District Court further by “zealously” insisting that that methodology, recently vindicated on appeal, must be abandoned. It sufficed, we think, for the State to submit for the record at every turn that “Defendants' objections and suggestions for modifications shall not be deemed a waiver of these Defendants’ right to appeal prior rulings and orders of this Court or appeal from the subsequent final Order setting forth the injunctive relief regarding legal access issues,” see, e. g., App. 221, 225, 231, 239, 243.