PHILLIPS, Circuit Judge:
The Commissioner of the South Carolina Department of Corrections and the members of the South Carolina Board of Corrections (collectively, SCDC or the state) appeal an order of the United States District Court for the District of South Carolina denying their motion for modification of a consent decree relating to the design, construction and operation of state prison facilities. Because we conclude that the court erred in declining to give the state any relief by modification of the decree, we vacate the order and remand with instructions for reconsideration of the motion.
I
This ease is a reprise of Plyler v. Evatt, 846 F.2d 208 (4th Cir.1988) (Plyler I), in which this court reversed the district court’s refusal to modify the consent decree in other particulars, and remanded with directions permanently to modify the decree as requested by the state. The general background leading to entry of the consent decree and the ensuing difficulties that arose in its implementation are extensively developed in the Plyler I opinion and need not be fully laid out again here. A brief summary of that background suffices.
The consent decree, signed by the parties in January 1985, settled a class action brought in 1982 by state prison inmates against state officials, claiming unconstitutional conditions of confinement in various state prison facilities. The decree effected a comprehensive settlement dealing with all facets of state prison administration as that bore upon conditions of confinement within the system. The gist of the settlement was an agreement by the state to improve prison conditions in specifically detailed ways in return for the inmates’ agreement to forego any claims they might have as a class for monetary or other relief because of the state’s maintenance of any provable unconstitutional conditions of confinement. The district court approved entry of the consent decree in March of 1986, after careful review of its details, and after specifically finding that it was knowingly and willingly entered into by the prisoner class and by responsible officials of the state. Because the class action was settled, the claim of unconstitutional conditions has never been adjudicated.
Although, as indicated, the decree swept broadly, its central provisions were those concerned with avoiding overcrowding and its attendant ills by laying down minimum space requirements principally expressed in terms of cell space per inmate. Timetables for achieving compliance in stages were provided. New and renovated facilities were subjected to specific space requirements for single and double cells. Old, existing facilities were subjected both to specific space requirements and to other, more flexible requirements designed to accommodate the practical constraints which their construction imposed on efforts to avoid overcrowding. Critical to this appeal were the restrictions for the then existing Womens’ Correctional Center (WCC), a facility built in 1973, which was one of three facilities that then housed women prisoners. At WCC the decree permitted the state originally to double-cell no more than one-half of the cells then being used for general population in five of the seven then existing forty-eight cell cottages and re[1323]*1323quired the state to construct additional cottage space to house 96 inmates by December 31, 1985.
Following entry of the consent decree, the state succeeded in meeting most of its obligations, with the notable exception of those pertaining to cell space. As to those central obligations, it has had difficulty from the outset due to an increase in prison population that quickly and substantially exceeded the projections upon which the state officials relied when the consent decree was signed. Indeed the state was already out of compliance with this critical aspect of the decree at the time of its entry in March of 1986.
Failure of the state to achieve compliance with various of the space requirements has led to a series of efforts by the inmate class to secure enforcement orders from the court, and responsive efforts by the state to obtain extensions of time to comply and, eventually, permanent modifications of the decree.
In July of 1986 the court granted a motion to enforce, and denied a countering motion to modify with respect to certain cell-space requirements. The state’s appeal from this order was dismissed as moot when the state achieved compliance with the requirements while the appeal was pending. Plyler v. Leeke, 804 F.2d 1251 (4th Cir.1986).
Again in May of 1987, when the inmate class sought an enforcement order, the state countered with a motion for permanent modification of the decree to permit it to continue non-complying double-celling of male inmates in certain newly constructed facilities. When the district court again refused to modify and instead ordered compliance, the state again appealed. This led to this court’s reversal in Plyler I.
The present appeal originated with a set of countering motions filed in early 1989 respecting the decree’s space requirements for women inmates at WCC. The inmate class had moved for an enforcement order; the state soon countered with a motion to modify the decree either temporarily through early January 1991, or permanently, to relieve it of its alleged noncompliance. It was undisputed at the time these motions were heard that the state was not in compliance as to the WCC requirements. Specifically, TV rooms in old cottages 1 through 7 were being used to house inmates although the decree specifically forbade this. More critically, all the cottages, 1 through 7, in existence when the decree was signed, were being fully double-celled, although the decree permitted double-cell-ing of no more than fifty percent of five of these facilities. Finally, inmates in two cottages constructed after the consent decree, cottages 8 and 9, were fully double-celled in cells of approximately 73 square feet, although the decree required that in these newly constructed facilities only single-celling should occur unless at least 100 square feet of cell space was provided.
In May 1989, the district court granted the state temporary relief from compliance with these provisions and set a hearing for August 1, 1989, to consider a more permanent solution. In the interval, the state was directed to submit alternative methods for achieving compliance. Following the requested submissions, the court conducted a hearing in September of 1989, and on February 15, 1990, entered an order which (1) denied the state’s motion for modification, either temporarily or permanently, and (2) required the state to “place the Womens’ Correctional Center in full compliance with the Consent Decree no later than April 2, 1990,” and provided that the state was “free to take whatever steps they deem advisable to gain such compliance.”
The state then took this appeal, and the district court’s order was stayed by this court pending the appeal.
II
A
The general principles guiding the district court’s consideration of the state’s motion to modify the consent decree and those guiding our review of its denial of that motion are well-settled. Plyler I, dealing with the same type issues raised in this case, recently restated those principles as [1324]*1324they apply to such issues, and we summarize them again here briefly.
Under its inherent equity powers, as now expressed in Fed.R.Civ.P. 60(b)(5), a district court may modify a judgment if “it is no longer equitable that the judgment should have prospective application.” Plyler I, 846 F.2d at 211. In exercise of that power, consent decrees such as that here in issue may be modified in appropriate cases on the basis of material changes in operative law or facts. Id. (citing Nelson v. Collins, 659 F.2d 420, 424 (4th Cir.1981) (en banc)). While underlying principles of finality and repose dictate great caution in modifying consent decrees on no other basis than material changes in law or fact, “the unique nature and demands of institutional reform litigation necessitate a more flexible approach to modification” than may be appropriate with respect to consent decrees between private parties. Id. at 212 (citing Ruiz v. Lynaugh, 811 F.2d 856, 860-61 (5th Cir.1987)) (citing New York State Ass’n for Retarded Children v. Carey, 706 F.2d 956, 970 (2d Cir.1983)). The “uniqueness” of institutional reform litigation lies in the fact that it is necessarily aimed at achieving “broad public policy objectives in a complex, ongoing fact situation,” with the consequence that consent decrees settling such litigation must be viewed as embodying “not so much peremptory commands to be obeyed in terms, as ... future-oriented plans designed to achieve [those] objectives.” Id. (citing Carey, 706 F.2d at 970 n. 17).
Prison reform litigation, and consent decrees settling such litigation, present particularly sensitive problems for courts asked by prison officials to modify decrees because of material changes in the “complex, ongoing fact situation” that inevitably underlies such decrees. For in this area there comes into play, as a counterweight to the general principle of restraint in modifying any judicial decree, the Supreme Court’s admonition of cautious deference to the judgment of government officials charged with the sensitive task of prison administration. Procunier v. Martinez, 416 U.S. 396, 404-05, 94 S.Ct. 1800, 1807-08, 40 L.Ed.2d 224 (1974); see also Duran v. Elrod, 760 F.2d 756, 759 (7th Cir.1985). In this particular realm of institutional reform, therefore, the general need for flexibility in considering requests for modification of complex decrees may be particularly acute.
In specific applications of these general equitable principles, more specific ones come into play. In general, modification should be granted only when the change of circumstances urged by the mov-ant was largely beyond that party’s control, and when compliance has been put beyond effective reach despite a good faith effort by the movant to comply. Nelson v. Collins, 659 F.2d 420, 424 (4th Cir.1981) (en banc). And in the end, the question whether equity requires modification is one committed to the district court’s discretion to be exercised through the familiar process of balancing the interests of the opposing parties and the public in light of relevant substantive principles. United States v. Swift & Co., 286 U.S. 106, 115-19, 52 S.Ct. 460, 462-64, 76 L.Ed. 999 (1932); Plyler I, 846 F.2d at 211.
In reviewing a district court’s ultimate decision on a motion for modification, we apply an abuse of discretion standard shaped by the above principles and in doing so we accept any predicate findings of fact made by the court that are not clearly erroneous. Plyler I, 846 F.2d at 212.
B
The modification requested here, like that requested in Plyler I, would, if granted, effectively have relieved the state, either permanently or at least through early 1991, of its obligations under the consent decree respecting, in this case, cell space requirements at WCC. The specific relief sought was permission to double-cell inmates in 165 general population cells of 73 to 83 square feet.1 As indicated, this was prohibited by the consent decree. The sole [1325]*1325basis for the requested relief by modification was, as in Plyler I, “unanticipated increases in prison population.”
Applying the equitable principles above summarized, the district court denied the motion for either temporary or permanent modification and instead ordered early compliance, specifically leaving the means of compliance to the state. We summarize the court’s reasoning, as reflected in its denial order.
On the fundamental question of the state’s noncompliance, a condition conceded by the state, the court found that it had long existed and that it had resulted in serious overcrowding at all the women’s facilities, including WCC. As of the date of the court’s order, the state was housing somewhere between 331 and 334 more inmates at WCC than allowed by the decree. On current best projections respecting the addition of new and shifted facilities, the state would remain out of compliance until January 1991. At that time, if a new 384-bed facility at Greenwood comes on line as planned, compliance could be achieved, but on present projections of inmate population trends, this would only last for two months. In the face of these facts, the state had not at the time of the court’s hearing appropriated any funds for additional facilities, though it was represented that the 1990 session of the General Assembly would consider an SCDC request for such funds.
The court next observed that in response to its direction that it suggest means for coming into compliance, the state had rejected as unacceptable the alternative of early release of a sufficient number of women inmates. The state had proposed instead that four 96-bed male work release centers then under construction be made available for women inmates, and that WCC be fully double-celled and that inmates be housed in TV rooms and the gymnasium at that facility. The court noted that this would apparently provide around 37 square feet of space for inmates double-celled in the existing cottage cells, around 23 square feet for inmates in the TV rooms, and would take the gymnasium out of use as a gymnasium.
The court then addressed the state’s contention that the requested modification was justified by the unanticipated, substantial increase in the female inmate population at WCC since the consent decree was entered. The court found that the increase indisputably had been substantial, to the point that the actual population projected for December 31, 1989, 790, was 60% greater than the projected figure, 493, on which the state acted in signing the consent decree in January 1985. Whether the increase fairly could be viewed as “unanticipated,” however, was thought doubtful in view of the early development of the trend. Irrespective of that, the court found that there had been a substantial change of operative facts which satisfied that predicate for relief by modification.
Turning to the balancing process, the court looked first to the interest of the plaintiff class. As to this, the court noted its understanding that Plyler I had held, with binding effect for this case, that the interest properly to be considered was not entitlement to the specific benefit of every term of the decree, but only the attainment of overall constitutional conditions of confinement — that being the “essence of their bargain.” The court recognized that double-celling is only one such condition, and that it is not, standing alone, unconstitutional. Hence, the court concluded that in order to find any harm to a cognizable interest of the plaintiff class resulting from modification of the decree, it would be necessary to find that the existing double-cell-ing, when considered in conjunction with all other conditions at WCC, would lead to unconstitutional conditions overall.
As to whether this could be found here, the court noted the difficulty that unconstitutionality overall had never been adjudicated, nor had an evidentiary basis for making the determination at this point been developed. The court observed, however, that the burden here was on the state as movant to show entitlement to relief, which would include proof that allowing double-celling by modifying the decree would not [1326]*1326tip the constitutional balance. Because the state had failed to carry that burden by failing to offer any evidence as to the general conditions at WCC, the court concluded that unconstitutionality overall had been established by the state’s failure to disprove it. From this it followed in the court’s reasoning that modification to permit continued double-celling would cause constitutional harm to the plaintiff class and necessarily deny it the "essence of the bargain” as that had been defined in Plyler I.
Turning to the state’s interest, the court recognized it as the obvious one of having the lawful sentences of its criminal courts fully carried out. Against this interest, however, the court thought it proper to weigh the state’s failure to make a good faith effort to comply with the consent decree provisions here in issue. This was found on the basis that despite unmistakable signs soon after the consent decree was entered that the state’s projections were heavily flawed, adequate responsive adjustments have never been made. Furthermore, the failure of the state to explain its opposition to early releases as a means of compliance, and to explain the effect on other provisions of the decree that would result from adoption of its proposal to convert male work release centers to female use, suggested to the court a present lack of good faith on the state’s part in seeking to comply with the decree.
The court then identified the principal interest of the public as lying in its protection from the risk of criminal conduct by the inmates presently confined in WCC. That required consideration by the court of the effect on this interest of the state’s being required to effect compliance through early releases of a sufficient number of inmates to make modification of the decree unnecessary. On this, the court found from the evidence of record concerning the number of inmates presently qualified for early release, that without any court order the state “could release enough nonviolent inmates from WCC to gain compliance with the Consent Decree without increasing the danger to the public in any significant degree.”
On balance, the court concluded that the competing interests of the parties and the public “overwhelmingly weigh against the [state].” For this reason the modification sought was deemed by the court not to be warranted.
C
Turning now to review of the district court’s ruling and its supporting rationale, we note at the outset that our earlier decision in Plyler I looms large over this case and substantially constrains its resolution through principles of res judicata and stare decisis. That case arose from comparable opposing efforts by the plaintiff class to hold the state to the letter of this consent decree and by the state to be relieved of the letter of the decree by having it modified to permit non-complying conditions to continue. While the specific provisions at issue in that case related to housing at other facilities within the prison system, a number of common or closely related issues of fact and law were necessarily and finally decided by this court in reversing the district court’s refusal to modify the decree in the respect there sought.
The district court in this case was of course well aware of the general prece-dential effect of our decision in Plyler I, but misperceived that effect in at least two critical respects. Because these are important to a proper resolution of this case, we identify them at this point, noting, that they cut in opposite directions.
First, as indicated, the district court read Plyler I as having identified the interest of (hence threatened harm to) the plaintiff class, for balancing purposes, as going only to “the attainment of constitutional prison conditions through the receipt of the ‘essence of their bargain’ made in the consent decree.” In other words, the court apparently thought that Plyler I allowed it to find cognizable harm to the plaintiff class resulting from modification of the decree only if it could determine that the modification would result in unconstitutional conditions — a matter that, as the court noted, had never been adjudicated.
[1327]*1327This was much too draconian a reading of Plyler 7’s holding on that point. As a moment’s reflection will show, so to read that decision would necessarily imply that the only legally enforceable obligation assumed by the state under the consent decree was that of ultimately achieving minimal constitutional prison standards. For the practical effect would be that every effort by the plaintiff class to enforce specific provisions of the decree could be effectively countered by a motion by the state to modify so long as the modification did not generate unconstitutional conditions overall. Substantively, this would do violence to the obvious intention of the parties that the decretal obligations assumed by the state were not confined to meeting minimal constitutional requirements. Procedurally, it would make necessary, as this case illustrates, a constitutional decision every time an effort was made either to enforce or modify the decree by judicial action. The district court avoided that procedural quagmire here by the awkward device of invoking the burden of proof against the state as movant for its failure to disprove unconstitutionality. But that simply cannot suffice in constitutional adjudication. Both the substantive and procedural implications of such a drastic reading of Plyler I are simply unacceptable.
Properly read, Plyler 7’s discussion of the nature of the plaintiff class’s interest for balancing purposes has much less drastic implications. The point being made in Plyler I was simply that in assessing the risk, and degree, of harm to the plaintiff class from the modification proposed there, attention should not be confined to the indisputable fact that modification would necessarily abrogate a specific benefit conferred by the decree. It must also take into account the extent to which the benefits still retained insured that the “broad public-policy objectives” of the decree were not at risk from the proposed modification. See Plyler I, 846 F.2d at 212-13.
The court’s other misperception of the effect of Plyler I was more critical, for it led the court into error in its crucial finding that the state must be charged with a lack of good faith in its failed efforts to comply with the WCC provisions of the decree. As indicated, this finding was central to the court’s conclusion that the state was not entitled to modification of the decree.
We believe that this lack of good faith finding, one of mixed law and fact, is foreclosed by our decision in Plyler I. The state’s good faith in its failed efforts to comply with the comparable cell-space provisions there in issue was a central issue in that case as it is here. See Plyler I, 846 F.2d at 213. In Plyler I, as here, the state based its request for modification on the simple fact that unanticipated increases in the prison population had made it impossible, despite its good faith efforts, to comply with the decree, and that this change in circumstances warranted relief by modification of the decree. In Plyler 7, as here, the district court had rejected the state’s critical claim that its failure to comply reflected a circumstance effectively beyond its control, and found instead, by necessary implication, that the state had not made a sufficiently good faith effort to warrant equitable relief. See id. On the state’s appeal, based on a detailed review of the state’s efforts to comply and the circumstances which had prevented compliance, we flatly rejected the district court’s crucial conclusion of a lack of good faith effort on the state’s part. We held instead that the state had made “good faith efforts to adhere to the consent decree,” and that the “direct cause” of the increase in prison population, hence of the state’s inability to comply with the decree, lay not in the state’s failure to control a controllable situation, but in the increased commission of crime, a factor that lay beyond its control. Id.
It is of course true that the cell-space provisions at issue here, hence the nature of the state’s specific obligations, are different from those at issue in Plyler I. Certainly it is theoretically possible that the state could have sought in good faith to comply with the provisions at issue in Plyler I, but not those at issue here. The district court, possibly seeking to avoid the precedential effect of the Plyler I decision, [1328]*1328indeed suggested that this was the ease— that while good faith may earlier have driven the state’s efforts, it had more recently faded and that this tended to explain the failure to comply as to the WCC provisions, if not those earlier at issue.
We do not believe that the precedential and preclusive effect of Plyler I properly can be avoided by that means. On any fair reading of that decision, it rejects as factually unsupported any finding that the state has not acted in good faith in its general efforts to comply with the cell-space provisions of the consent decree in the face of unanticipated increases in prison population. We think that any decision which purported now to find a lack of good faith on the part of the state in its efforts to comply with the specific WCC cell-space provisions would necessarily conflict to a significant degree with the broad findings of good faith made in Plyler I. That, of course, cannot be allowed, and we therefore must reject on this basis2 the district court’s crucial finding of a lack of good faith on the state’s part here.
With that crucial finding rejected, the district court’s balancing of interests, in which the state’s lack of good faith played a critical part, is also undercut. Here again, we think that Plyler I’s balancing analysis, in which the state’s essential good faith is accepted, controls decision here. That analysis led to the conclusion in Plyler I that modification of the decree was warranted. We think the same analysis compels a comparable conclusion here.
In sum, we read Plyler I as having established as the law of this case the broad proposition that the general increase in all segments of the prison population in South Carolina since entry of the consent decree must be considered to have been reasonably unanticipated by the state, and that the state’s failure to achieve compliance with the various cell-space provisions of the consent decree must in fairness be attributed to this unanticipated and uncontrollable development rather than to any lack of good faith effort on the state’s part. From this necessarily follows the further proposition, which we also must accept as the law of this case, that strict compliance, on schedule, with the various cell-space provisions of the consent decree has now been revealed to be unachievable for reasons beyond the state’s control and that those reasons warrant appropriate equitable relief upon proper request by the state. This means that general entitlement to some measure of equitable relief from these particular provisions of the consent decree should not be subject to continuous relit-igation in this ongoing matter; general entitlement has been settled, barring drastic new developments not yet shown. But the question of what may be appropriate equitable relief from particular cell-space provisions of course must remain a matter for the discretionary judgment of the injunction court. Obviously, Plyler I did not establish that appropriate relief is in every case the total, permanent abrogation of a specific obligation that was found appropriate in that case.3 Instead specific relief must be shaped by the general equitable considerations peculiar to institutional reform litigation that we noted in Part II-A, [1329]*1329and of course by practical considerations related to the particular provision at issue and the ability of the state to achieve compliance with any modified obligation that can be constructed.
Ill
It remains now to identify the precise error we find in the court’s denial of the state’s motion for modification, and to shape a proper mandate for correction of the error on remand.
The specific error we find is the court’s refusal to give any relief from the cell-space requirements at issue. As indicated, the state sought either permanent abrogation of the particular cell-space provisions which prevented double-celling in 165 general population cells at WCC, or temporary relief from the provisions through January 1991. The district court declined to give either form of relief and instead ordered immediate, strict compliance by whatever means the state chose.
As we have held, Plyler I effectively established the state’s entitlement to some form of appropriate relief from strict compliance with these cell-space provisions. On remand, the district court must therefore set its hand to devising that form of relief which will best insure continued progress toward achievement of the broad public policy objective of relieving overcrowding that underlies the cell-space requirements of the decree, while accommodating the present impossibility of achieving that objective by the originally intended means of constructing additional facilities on a specific timetable.
In seeking the appropriate accommodation, the court may of course continue to consider the alternative of early release of inmates as a device for achieving in whole or part the ultimate objectives at which the cell-space requirements are aimed. But for reasons specifically addressed in Plyler I, see 846 F.2d at 213-14, that alternative must be considered the accommodation of last resort, in view of the essentially unmeasurable but profound public interests which it implicates. See id. We note, however, that nothing said on the subject of early releases in Plyler I rules out the possibility that this draconian last alternative might properly be employed by the court on a sufficient showing of newly arisen bad faith on the state’s part.
Obviously, the court on remand must also take into account any relevant developments in the interval since its order was entered. Of particular relevance would be any legislation addressing the general problem and the advent or imminent prospect of new facilities becoming available.4
In conclusion, we offer these observations for general guidance of the district court in the ongoing process of monitoring compliance with this consent decree in response to recurring and conflicting requests for cell-space relief by the parties. It is obvious to this court from its review of a succession of orders by the district court respecting the state’s failure to achieve compliance with various cell-space requirements, that the district court has been and remains concerned with the bona fides of the state’s efforts. This obviously has led the district court to feel that only by ordering strict compliance with these requirements (and denying any sought modification) may the objectives of the consent decree and fairness to the plaintiff class be achieved. In short, it is manifest that the district court has been convinced that the state is deliberately dragging its feet in this matter and has declined to do what it could and should in equity do to achieve compliance with the critical cell-space requirements of the consent decree. What must be accepted now, however, is that this view has been forcefully contested and defended in this court by skilled and dedicated counsel and by this court rejected. That the rejection was by a split panel [1330]*1330decision attests that the question was a close and difficult one whose resolution implied no reproach to the district court’s professional judgment, but simply the imposition of a conflicting judgment by a reviewing court. Nevertheless, as our decision in this case has emphasized, this basic view of the lack of bona fides of the state’s efforts has now been authoritatively rejected and the contrary view imposed as the law of this case.
This means that the guiding principle for the district court from this point on in dealing with comparable issues must be to accept the necessity to make appropriate accommodations to what has now been determined to be a situation brought on, at least to this point in time, by factors beyond the state’s effective control. This of course assumes that there shall not have been any supervening events after the date of this decision which draw in issue the continuing bona fides of the state. And it assumes the continuing obligation of the court to make only those accommodations which do not jeopardize achievement of the consent decree’s ultimate objectives, including in particular those related to the provision of humane living space. In sum, the watchword must be flexibility in keeping such pressure on as is necessary to insure continued good faith efforts to comply while accommodating the reality of present inability to meet particular cell-space requirements.
For the reasons given, we vacate the district court’s order denying the state’s motion for modification, and remand for reconsideration of the motion in accordance with this opinion.
VACATED AND REMANDED.