---

Justice Breyer, with whom Justice Stevens joins, dissenting.

The Prison Litigation Reform Act of 1995 (PLRA) says that “any party or intervener” may move to terminate any “prospective relief” previously granted by the court, 18 U.S.C. § 3626(b)(1) (1994 ed., Supp. IV), and that the court shall terminate (or modify) that relief unless it is “necessary to correct a current and ongoing violation of [a] Federal right, extends no further than necessary to correct the violation … [and is] the least intrusive means” to do so. 18 U.S.C. § 3626(b)(3).

We here consider a related procedural provision of the PLRA. It says that “[a]ny motion to modify or terminate prospective relief … shall operate as a stay” of that prospective relief “during the period” beginning (no later than) the 90th day after the filing of the motion and ending when the motion is decided. §3626(e)(2). This provision means approximately the following: Suppose that a district court, in 1980, had entered an injunction governing present and future prison conditions. Suppose further that in 1996 a party filed a motion under the PLRA asking the court to terminate (or to modify) the 1980 injunction. That district court would have no more than 90 days to decide whether to grant the motion. After those 90 days, the 1980 injunction would terminate automatically–regaining life only if, when, and to the extent that the judge eventually decided to deny the PLRA motion.

The majority interprets the words “shall operate as a stay” to mean, in terms of my example, that the 1980 injunction must become ineffective after the 90th day, no matter what. The Solicitor General, however, believes that the view adopted by the majority interpretation is too rigid and calls into doubt the constitutionality of the provision. He argues that the statute is silent as to whether the district court can modify or suspend the operation of the automatic stay. He would find in that silence sufficient authority for the court to create an exception to the 90-day time limit where circumstances make it necessary to do so. As so read, the statute would neither displace the courts’ traditional equitable authority nor raise significant constitutional difficulties. See Califano v. Yamasaki, 442 U.S. 682, 705 (1979) (only “clearest” congressional “command” displaces courts’ traditional equity powers); Edward J. DeBartolo Corp. v. Florida Gulf Coast Building & Constr. Trades Council, 485 U.S. 568, 575 (1988) (the Court will construe a statute to avoid constitutional problems “un-
less such construction is plainly contrary to the intent of Congress”).

I agree with the Solicitor General and believe we should adopt that “ ‘reasonable construction’ ” of the statute. Ibid. (quoting Hooper v. California, 155 U.S. 648, 657 (1895), stating “ ‘every reasonable construction must be resorted to, in order to save a statute from unconstitutionality’ ”).

At the outset, one must understand why a more flexible interpretation of the statute might be needed. To do so, one must keep in mind the extreme circumstances that at least some prison litigation originally sought to correct, the complexity of the resulting judicial decrees, and the potential difficulties arising out of the subsequent need to review those decrees in order to make certain they follow Congress’ PLRA directives. A hypothetical example based on actual circumstances may help.

In January 1979, a Federal District Court made 81 factual findings describing extremely poor–indeed “barbaric and shocking”–prison conditions in the Commonwealth of Puerto Rico. Morales Feliciano v. Romero Barcelo, 497 F. Supp. 14, 32 (PR 1979). These conditions included prisons typically operating with twice the number of prisoners they were designed to hold; inmates living in 16 square feet of space (i.e., only 4 feet by 4 feet); inmates without medical care, without psychiatric care, without beds, without mattresses, without hot water, without soap or towels or toothbrushes or underwear; food prepared on a budget of $1.50 per day and “tons of food … destroyed because of … rats, vermin, worms, and spoilage”; “no working toilets or showers,” “urinals [that] flush into the sinks,” “plumbing systems … in a state of collapse,” and a “stench” that was “omnipresent”; “exposed wiring … no fire extinguisher, … [and] poor ventilation”; “calabozos,” or dungeons, “like cages with bars on the top” or with two slits in a steel door opening onto a central corridor, the floors of which were “covered with raw sewage” and which contained prisoners with severe mental illnesses, “caged like wild animals,” sometimes for months; areas of a prison where mentally ill inmates were “kept in cells naked, without beds, without mattresses, without any private possessions, and most of them without toilets that work and without drinking water.” Id., at 20—23, 26—27, 29, 32. These conditions had led to epidemics of communicable diseases, untreated mental illness, suicides, and murders. Id., at 32.

The District Court held that these conditions amounted to constitutionally forbidden “cruel and unusual punishment.” Id., at 33—36. It entered 30 specific orders designed to produce constitutionally mandated improvement by requiring the prison system to, for example, screen food handlers for communicable diseases, close the “calabozos,” move mentally ill patients to hospitals, fix broken plumbing, and provide at least 35 square feet (i.e., 5 feet by 7 feet) of living space to each prisoner. Id., at 39—41.

The very pervasiveness and seriousness of the conditions described in the court’s opinion made those conditions difficult to cure quickly. Over the next decade, the District Court entered further orders embodied in 15 published opinions, affecting 21 prison institutions. These orders concerned, inter alia, overcrowding, security, disciplinary proceedings, prisoner classification, rehabilitation, parole, and drug addiction treatment. Not surprisingly, the related proceedings involved extensive evidence and argument consuming thousands of pages of transcript. See Morales Feliciano v. Romero Barcelo, 672 F. Supp. 591, 595 (PR 1986). Their implementation involved the services of two monitors, two assistants, and a Special Master. Along the way, the court documented a degree of “administrative chaos” in the prison system, Morales Feliciano v. Hernandez Colon, 697 F. Supp. 37, 44 (PR 1988), and entered findings of contempt of court against the Commonwealth, followed by the assessment and collection of more than $74 million in fines. See Morales Feliciano v. Hernandez Colon, 775 F. Supp. 487, 488 and n. 2 (PR 1991).

Prison conditions subsequently have improved in some respects. Morales Feliciano v. Rossello Gonzalez, 13 F. Supp. 2d 151, 179 (PR 1998). I express no opinion as to whether, or which of, the earlier orders are still needed. But my brief summary of the litigation should illustrate the potential difficulties involved in making the determination of continuing necessity required by the PLRA. Where prison litigation is as complex as the litigation I have just described, it may prove difficult for a district court to reach a fair and accurate decision about which orders remain necessary, and are the “least intrusive means” available, to prevent or correct a continuing violation of federal law. The orders, which were needed to resolve serious constitutional problems and may still be needed where compliance has not yet been assured, are complex, interrelated, and applicable to many different institutions. Ninety days might not provide sufficient time to ascertain the views of several different parties, including monitors, to allow them to present evidence, and to permit each to respond to the arguments and evidence of the others.

It is at least possible, then, that the statute, as the majority reads it, would sometimes terminate a complex system of orders entered over a period of years by a court familiar with the local problem–perhaps only to reinstate those orders later, when the termination motion can be decided. Such an automatic termination could leave constitutionally prohibited conditions unremedied, at least temporarily. Alternatively, the threat of termination could lead a district court to abbreviate proceedings that fairness would otherwise demand. At a minimum, the mandatory automatic stay would provide a recipe for uncertainty, as complex judicial orders that have long governed the administration of particular prison systems suddenly turn off, then (perhaps selectively) back on. So read, the statute directly interferes with a court’s exercise of its traditional equitable authority, rendering temporarily ineffective pre-existing remedies aimed at correcting past, and perhaps ongoing, violations of the Constitution. That interpretation, as the majority itself concedes, might give rise to serious constitutional problems. Ante, at 21.

The Solicitor General’s more flexible reading of the statute avoids all these problems. He notes that the relevant language says that the motion to modify or terminate prospective relief “shall operate as a stay” after a period of 30 days, extendable for “good cause” to 90 days. 18 U.S.C. § 3626(e)(2); see also Brief for United States 12. The language says nothing, however, about the district court’s power to modify or suspend the operation of the “stay.” In the Solicitor General’s view, the “stay” would determine the legal status quo; but the district court would retain its traditional equitable power to change that status quo once the party seeking the modification or suspension of the operation of the stay demonstrates that the stay “would cause irreparable injury, that the termination motion is likely to be defeated, and that the merits of the motion cannot be resolved before the automatic stay takes effect.” Ibid. Where this is shown, the “court has discretion to suspend the automatic stay and require prison officials to comply with outstanding court orders until the court resolves the termination motion on the merits,” id., at 12—13, subject to immediate appellate review, 18 U.S.C. § 3626(e)(4).

Is this interpretation a “reasonable construction” of the statute? Edward J. DeBartolo Corp., 485 U.S., at 575. I note first that the statutory language is open to the Solicitor General’s interpretation. A district court ordinarily can stay the operation of a judicial order (such as a stay or injunction), see Scripps-Howard Radio, Inc. v. FCC, 316 U.S. 4, 9—10, and n. 4 (1942), when a party demonstrates the need to do so in accordance with traditional equitable criteria (irreparable injury, likelihood of success on the merits, and a balancing of possible harms to the parties and the public, see Doran v. Salem Inn, Inc., 422 U.S. 922, 931 (1975); Yakus v. United States, 321 U.S. 414, 440 (1944)). There is no logical inconsistency in saying both (1) a motion (to terminate) “shall operate as a stay,” and (2) the court retains the power to modify or delay the operation of the stay in appropriate circumstances. The statutory language says nothing about this last-mentioned power. It is silent. It does not direct the district court to leave the stay in place come what may.

Nor does this more flexible interpretation deprive the procedural provision of meaning. The filing of the motion to terminate prospective relief will still, after a certain period, operate as a stay without further action by the court. Thus, the motion automatically changes the status quo and imposes upon the party wishing to suspend the automatic stay the burden of demonstrating strong, special reasons for doing so. The word “automatic” in the various subsection titles does not prove the contrary, for that word often means self-starting, not unstoppable. See Websters Third New International Dictionary 148 (1993). Indeed, the Bankruptcy Act uses the words “automatic stay” to describe a provision stating that “a petition filed … operates as a stay” of certain other judicial proceedings–despite the fact that a later portion of that same provision makes clear that under certain circumstances the bankruptcy court may terminate, annul, or modify the stay. 11 U.S.C. § 362(d); see also 143 Cong. Rec. S12269 (Nov. 9, 1997) (statement of Sen. Abraham) (explaining that §3626(e)(2) was modeled after the Bankruptcy Act provision). And the Poultry Producers Financial Protection Act of 1987 specifies that a court of appeals decree affirming an order of the Secretary of Agriculture “shall operate as an injunction” restraining the “live poultry dealer” from violating that order, 7 U.S.C. § 228b—3(g); yet it appears that no one has ever suggested that the court of appeals lacks the power to modify that “injunction” where appropriate. Moreover, the change in the legal status quo that the automatic stay would bring about, and the need to demonstrate a special need to lift the stay (according to traditional equitable criteria), mean that the stay would remain in effect in all but highly unusual cases.

In addition, the surrounding procedural provisions are most naturally read as favoring the flexible interpretation. The immediately preceding provision requires the court to rule “promptly” upon the motion to terminate and says that “[m]andamus shall lie to remedy any failure to issue a prompt ruling.” 18 U.S.C. § 3626(e)(1). If a motion to terminate takes effect automatically through the “stay” after 30 or 90 days, it is difficult to understand what purpose would be served by providing for mandamus–a procedure that itself (in so complicated a matter) could take several weeks. But if the automatic stay might be modified or lifted in an unusual case, providing for mandamus makes considerable sense. It guarantees that an appellate court will make certain that unusual circumstances do in fact justify any such modification or lifting of the stay. A later provision that provides for immediate appeal of any order “staying, suspending, delaying or barring the operation of the automatic stay” can be read as providing for similar appellate review for similar reasons. §3626(e)(4).

Further, the legislative history is neutral, for it is silent on this issue. Yet there is relevant judicial precedent. That precedent does not read statutory silence as denying judges authority to exercise their traditional equitable powers. Rather, it reads statutory silence as authorizing the exercise of those powers. This Court has said, for example, that “[o]ne thing is clear. Where Congress wished to deprive the courts of this historic power, it knew how to use apt words–only once has it done so and in a statute born of the exigencies of war.” Scripps-Howard, supra, at 17. Compare Lockerty v. Phillips, 319 U.S. 182, 186—187 (1943) (finding that courts were deprived of equity powers where the statute explicitly removed jurisdiction), with Scripps-Howard, 316 U.S., at 8—10 (refusing to read silence as depriving courts of their historic equity power), and Califano, 442 U.S., at 705—706 (same). These cases recognize the importance of permitting courts in equity cases to tailor relief, and related relief procedure, to the exigencies of particular cases and individual circumstances. In doing so, they recognize the fact that in certain circumstances justice requires the flexibility necessary to treat different cases differently–the rationale that underlies equity itself. Cf. Hecht Co. v. Bowles, 321 U.S. 321, 329 (1944) (“The essence of equity jurisdiction has been the power of the Chancellor to do equity and to mould each decree to the necessities of the particular case”).

Finally, the more flexible interpretation is consistent with Congress’ purposes as revealed in the statute. Those purposes include the avoidance of new judicial relief that is overly broad or no longer necessary and the reassessment of pre-existing relief to bring it into conformity with these standards. But Congress has simultaneously expressed its intent to maintain relief that is narrowly drawn and necessary to end unconstitutional practices. See 18 U.S.C. § 3626(a)(1), (a)(2), (b)(3). The statute, as flexibly interpreted, risks interfering with the first set of objectives only to the extent that the speedy appellate review provided in the statute fails to control district court error. The same interpretation avoids the improper provisional termination of relief that is constitutionally necessary. The risk of an occasional small additional delay seems a comparatively small price to pay (in terms of the statute’s entire set of purposes) to avoid the serious constitutional problems that accompany the majority’s more rigid interpretation.

The upshot is a statute that, when read in light of its language, structure, purpose, and history, is open to an interpretation that would allow a court to modify or suspend the automatic stay when a party, in accordance with traditional equitable criteria, has demonstrated a need for such an exception. That interpretation reflects this Court’s historic reluctance to read a statute as depriving courts of their traditional equitable powers. It also avoids constitutional difficulties that might arise in unusual cases.

I do not argue that this interpretation reflects the most natural reading of the statute’s language. Nor do I assert that each individual legislator would have endorsed that reading at the time. But such an interpretation is a reasonable construction of the statute. That reading harmonizes the statute’s language with other basic legal principles, including constitutional principles. And, in doing so, it better fits the full set of legislative objectives embodied in the statute than does the more rigid reading that the majority adopts.

For these reasons, I believe that the Solicitor General’s more flexible reading is the proper reading of the statute before us. I would consequently vacate the decision of the Court of Appeals and remand this action for further proceedings.