The Commissioner of the South Carolina Department of Corrections and the members of the South Carolina Board of Corrections (Department of Corrections) appeal from the denial of a motion for modification of a consent decree regarding correctional facilities. The Department of Corrections also appeals from a recent court-ordered plan requiring early release of approximately 700 prisoners. We vacate the orders and remand for modification of the consent decree consistent with this opinion.
I.
This action was initiated in 1982 when a class of inmates filed suit against the Department of Corrections under 42 U.S.C.A. § 1983 (West 1981), complaining of overcrowded conditions in state prisons. The State of South Carolina, acting through the General Assembly, authorized negotiations, expressing intent “by legislation and appropriations to implement a reasonable settlement of the issues.” H.Cong.Res. No. 3054,1983, S.C. House J. 2837. The General Assembly approved a general agreement reached by the parties in 1984 and simultaneously appropriated funds for capital improvements. Appropriations Act, Act No. 512, Part III, § IV, 1984 S.C. Acts 2176, 3107. A proposed settlement agreement was signed in January 1985 by then Commissioner William D. Leeke, all active members of the Board of Corrections, a representative of the Attorney General, counsel for the inmates, and 26 representatives of the inmate class. The Department of Corrections began immediate implementation of the settlement agreement. Subsequently, the district court orally approved the agreement in November 1985 and by written consent decree in March 1986. Among many other things, the consent decree placed certain requirements on new construction and renovation. It provided that new cells housing a single inmate contain at least 50 square feet and new cells housing two inmates contain at least 100 square feet. The decree further mandated increased staffing, improved medical care, and expansion of educational, vocational *210and recreational programs. Implementation of the provisions was scheduled to take place in stages over a period of five years.
After the agreement was reached, the Department of Corrections began to experience an unanticipated growth in inmate population which made it impossible to comply with some of the housing provisions of the consent decree. As a result, in July 1986 the Department of Corrections sought a temporary delay in compliance pending implementation of a State early-release program. The district court refused to grant an extension and ordered compliance. Although the Department of Corrections appealed, it achieved compliance in the interim, and the issue was mooted. Plyler v. Leeke, 804 F.2d 1251 (4th Cir.1986) (per curiam).
In May 1987, because of the unanticipated increase in inmate population, the Department of Corrections moved for a permanent modification of the consent decree to allow double-celling in five new facilities. Prior to the motion reaching the district court for disposition, a mediator and a United States Magistrate conducted hearings and made recommendations. The mediator recommended that temporary double-celling be allowed in 50% of the new cells. The magistrate recommended that double-celling be permitted in new cells for two years. Rejecting these recommendations in January 1988, the district court refused the modification request and ordered immediate compliance through early-release programs or other appropriate means. The district court then required the Department of Corrections to submit a plan which would have provided for the early release of approximately 700 inmates over the next three months. The district court orders were stayed pending this appeal.
II.
Under the consent decree, inmates in the general population in new facilities who are confined to a cell less than 12 hours per day may be double-celled if the cells contain at least 100 square feet. The Department of Corrections seeks a permanent modification of this provision to allow double-celling at five new medium security facilities in cells which provide 69 to 73 square feet, depending on the facility. Two of these facilities, Lieber and McCormick, are now open, and the others, Broad River, Allendale and Marlboro, are in various stages of construction. Construction schedules call for Broad River to open in April 1988, and Allendale and Marlboro to open in early 1989.
Lieber contains 504 cells of 73 square feet each, and McCormick and Broad River have 504 cells of 69 square feet each.1 The plans for Allendale and Marlboro were altered in May 1987 to provide for 296 cells of 69 square feet each and 208 cells of 100 square feet each.2 The design plans for these five new facilities are based on the plans used for construction of the Federal Correctional Institution in Phoenix, Arizona which represents state of the art in correctional facilities. The cells are individual rooms constructed of painted concrete block with tile floors, each with a window and door. Each room is furnished with a sink and toilet, lockers, a desk and chair, and bunk beds. The inmates are allowed to have personal items, such as radios and televisions. They are confined to their rooms only from 11:30 p.m. to 6:00 a.m. All cells are centrally heated and air-conditioned.
These new facilities also provide a variety of amenities. For example, Lieber offers vocational programs in plumbing, carpentry, horticulture and automotive repair. The academic programs include high school *211equivalency and college classes. The prison industry employs inmates in automotive refurbishing. A variety of recreational activities is provided, including sports, crafts, and music. Dayrooms, visiting areas, and a cafeteria dining room are also provided.
III.
The consent decree allows for modification by mutual and joint petition of the parties, and “[a]ny disputed petition for modification shall be reviewed by the Court under the applicable law pertaining to modification of Consent Decrees.” The consent decree further provides a procedure by which the Department of Corrections may obtain variances from the space allotments in new construction not specifically addressed in the consent decree, such as Broad River, Allendale and Marlboro. In the event of a dispute, the court is empowered to “make a determination as to the reasonableness of or necessity for said variance, in light of but not limited to the requirements of this Decree and the totality of the conditions, and shall determine whether or not said variance shall be permitted.”
The Department of Corrections seeks modification because of an unanticipated increase in prison population. As specified in the consent decree, it anticipated an average net increase of no less than 30, but no more than 50 prisoners per month from 1985 to 1990. In the past, official predictions had been fairly accurate. However, in the past several years actual increases in the number of inmates housed in Department of Corrections facilities have substantially exceeded the estimates. Statistics stipulated to by the parties show that monthly increases for 1985 reached a high of 212 in October with a yearly average of 74 per month. In 1986, highs of 233 occurred in March and September with a yearly average of 84 per month. The high for 1987 was 158 in October and the yearly average was 59 per month.
Pursuant to the decree the parties agreed that if the predictions proved to be inaccurate, “the Court shall order immediate relief, which may include population reductions, release or transfer of prisoners ... or other appropriate relief.” We find that the appropriate remedy is modification of the decree to allow double-celling in the five new facilities.
IV.
Generally, Federal Rule of Civil Procedure 60(b)(5) provides that the court may modify an order if “it is no longer equitable that the judgment should have prospective application.” Under Nelson v. Collins, 659 F.2d 420, 424 (4th Cir.1981) (en banc), a consent decree may be modified in response to material changes in operative law or facts. As found by the district court, the unanticipated increase in population clearly is a change in operative facts which meets this predicate for modification. In addition, the court must balance the competing interests of the prisoners, the Department of Corrections, and the public. “[T]he interests of the prisoners in immediate and strict enforcement of the consent decree” must be weighed against the Department of Corrections’ interest “in the orderly administration of the corrections system” and the public interest “in having lawful sentences carried out and in not having parolees put at large without sufficient supervision.” Plyler v. Leeke, No. 86-7654, slip op. at 7 (4th Cir. Nov. 12, 1986) [804 F.2d 1251 (table) ] (per curiam).
The dissent maintains that the State must demonstrate oppression and that modification should only be allowed “upon a strong showing of near compulsion to grant relief.” Infra at 216 (citing Holiday Inns, Inc. v. Holiday Inn, 645 F.2d 239, 240 (4th Cir.), cert. denied, 454 U.S. 1053, 102 S.Ct. 597, 70 L.Ed.2d 588 (1981); United States v. Swift & Co., 286 U.S. 106, 119, 52 S.Ct. 460, 464, 76 L.Ed. 999 (1932); Humble Oil & Ref. Co. v. American Oil Co., 405 F.2d 803, 813 (8th Cir.), cert. denied, 395 U.S. 905, 89 S.Ct. 1745, 23 L.Ed.2d 218 (1969)).3 Although these cases may *212set a strict standard for modification of consent decrees between private parties, this standard is inappropriate in institutional reform litigation for “the unique nature and demands of institutional reform litigation necessitate a more flexible approach to modification.” Ruiz v. Lynaugh, 811 F.2d 856, 860-61 (5th Cir.1987) (footnote omitted) (citing New York State Ass’n for Retarded Children, Inc. v. Carey, 706 F.2d 956, 970 (2d Cir.), cert. denied, 464 U.S. 915, 104 S.Ct. 277, 78 L.Ed.2d 257 (1983)).
The Constitution prohibits cruel and unusual punishment, but it does not specifically address prison conditions. For this reason, federal courts have traditionally adopted a policy of judicial restraint in the problematic area of prison administration. Procunier v. Martinez, 416 U.s. 396, 404-405, 94 S.Ct. 1800, 1807, 40 L.Ed.2d 224 (1974); Duran v. Elrod, 760 F.2d 756, 759 (7th Cir.1985). Federal courts do intervene when prison conditions offend fundamental constitutional guarantees, Procunier, 416 U.S. at 405-06, 94 S.Ct. at 1807, but "the threat and constitutional value that occasions the intervention can never be defined with great precision.... [Therefore,] revision is justified if the remedy is not working effectively or is unnecessarily burdensome." Carey, 706 F.2d at 970 (quoting Fiss, The Supreme Court-1978 Term-Foreword: The Forms of Justice, 93 Harv.L.Rev. 1, 49 (1979)).
The district court here followed the flexible approaches utilized by this court in Nelson v. Collins and the earlier appeal in Plyler v. Leeke, finding that:
The need for the decree far outweighs the harm that will result to the defendants or the public if the decree is not modified. This court remains convinced that the public interest is best served by requiring the State of South Carolina to perform the promises it made to its citizens and others in this matter.
Plyler v. Leeke, No. 3:82-0876-2, slip op. at 12 (D.S.C. Jan. 11,1988). Under the proper standard of judicial review, the district court’s balancing of the competing interests will not be set aside except for an abuse of discretion. Plyler v. Leeke, No. 86-7654, slip op. at 11 (4th Cir. Nov. 12, 1986) [804 F.2d 1251 (table) ] (per curiam); Duran, 760 F.2d at 762. In assessing the district court’s exercise of discretion, its findings of fact are subject to the clearly erroneous standard. Fed.R.Civ.P. 52(a). We find that the district court clearly erred in assessing the degree of potential harm to the inmates as contrasted with the risks to the public and it abused its discretion in denying the current request for modification.
A.
First, the district court erred in assessing the right of the prisoners to the benefits of the settlement. As the Second Circuit recognized in Carey, consent decrees such as this in institutional reform cases "are not so much peremptory commands to be obeyed in terms, as they are future-oriented plans designed to achieve broad public policy objectives in a complex, ongoing fact situation." 706 F.2d at 970 n. 17 (quoting Chayes, The Supreme Court-1981 Term -Foreword: Public Law Litigation and the Burger Court, 96 Harv.L.Rev. 4, 56 (1982)). The court here failed to recognize that the central goal of the decree is to provide constitutional prison conditions, and instead focused inappropriately on the double-ceiling provision.
Although double-celling will be contrary to a specific term of the consent decree, the prisoners have received the essence of their bargain. Not only have all of the many terms of the agreement, except double-cell-*213ing in five new facilities, been met, but also the general conditions of confinement now not only meet, but exceed constitutional requirements. In the balance, compelling the State to achieve compliance through the early release of massive numbers of inmates would create substantial dangers which are unjustified in view of the State’s good faith efforts to adhere to the consent decree.
B.
The State has embarked on an aggressive agenda of new prison construction, expending more than one hundred million dollars on capital improvements over the past few years. The State forged ahead with construction of Lieber even before a final settlement was reached. And, the facilities at Broad River, Allendale and Marlboro are under construction even though not specifically required by the agreed upon construction and renovation schedule. The General Assembly appropriated $155,000,000.00 in capital improvement financing for the Department of Corrections for 1984 to 1987 which includes the new construction costs at an average of $41,000.00 per cell. Under no circumstances can it be argued that the State has been miserly or dilatory.
We accept the district court’s finding that the increases in inmate population were, to some extent, within the State’s control, but we disagree with the dissent that we must accept the implicit finding that the State has not made a good faith effort to comply with the decree. Even accepting the finding that the increases were in part due to criminal laws enacted after the consent decree and to new parole policies, this finding is not conclusive on the issue of good faith. The district court failed to distinguish the State’s indirect control from the direct cause of the increases — the commission of crimes. Further, enactment of stricter criminal laws should not be considered bad faith. It is not disputed that the State is meeting its responsibility to provide constitutional prison conditions. It need not allow those convicted of crimes to go unpunished in order to comply with a single, arbitrary provision governing the size of prison cells.
C.
While proceeding with new construction, the State also instituted early-release programs to alleviate prison overcrowding. Prison Overcrowding Powers Act, S.C.Code Ann. §§ 24-3-1110, et seq. (Law. Co-op. Supp.1987) (also known as the Emergency Powers Act; EPA I, EPA II); Supervised Furlough Program, S.C.Code Ann. §§ 24-13-710, et seq. (Law. Co-op. Supp.1987). Approximately 7,000 inmates were released under these programs from September 1983 to August 1987. Of these 7,000, 18.8% returned to prison after the commission of new crimes, 1.3% for violent crimes and 17.5% for nonviolent crimes. As the district court found, the recidivism rate for inmates released under these programs was lower than for inmates released through the normal parole process. Based on these statistics, the district court was “not persuaded that public safety would be substantially lessened by the employment of these early release mechanisms.” Slip op. at 10.
However, the district court failed to consider additional evidence which clearly demonstrates that the continued effectiveness of these programs has been impaired because their extensive use has depleted the pool of inmates who can be released with a relative degree of safety to the public.4 As of August 1, 1987, shortly prior to the evidentiary hearing held by the magistrate, 880 inmates were statutorily eligible for early release under EPA II. Of these 880 inmates, 30 were classified as low risk, 157 as moderate risk, and 693 as high risk. Low risk inmates have a 7.6% failure rate upon release.5 In comparison, the failure *214rates for moderate and high risk inmates are 22.6% and 44.7%, respectively.
The district court failed to consider other potential dangers from the early release of high risk inmates in addition to the threat to public safety. Under the court-ordered plan inmates would be released from minimum security prisons, and medium security inmates then would be transferred to the minimum security facilities. This would create a potential danger to the remaining minimum security inmates and Department of Corrections personnel. There would also be an increased risk to the general public due to a greater potential for escapes from the minimum security facilities, some of which are not enclosed by a single fence.
These dangers far outweigh any imposition on the inmates from double-celling. It is undisputed that double-celling in these modern, air-conditioned facilities does not fall below constitutional standards. See Rhodes v. Chapman, 452 U.S. 337, 101 S.Ct. 2392, 69 L.Ed.2d 59 (1981). Interestingly, the consent agreement allows continued partial double-celling at older Department of Corrections facilities such as Central Correctional Institution, Perry Correctional Institution, and Kirkland Correctional Institution in cells of 56.36, 60.5, and 67.4 square feet, respectively. The total environment of these new facilities is without question superior to that of the older facilities.
D.
The district court found that prison overcrowding creates the potential for greater violence and makes proper management more difficult, relying on the testimony of Deputy Commissioner Catoe that “there’s a generally accepted position on the part of people who run institutions that as an institution gets bigger, as the inmate population increases, the management of that institution becomes more difficult.” The court’s finding based on such a general statement is clearly erroneous when measured against contradictory objective evidence of the actual incidences of violence and Commissioner Catoe’s inconsistent opinion that the statistics did not show a “disproportionate increase in violence.” See Anderson v. Bessemer City, 470 U.S. 564, 575, 105 S.Ct. 1504, 1512, 84 L.Ed.2d 518 (1985) (“Documents or objective evidence may contradict the witness’ story; or the story itself may be so internally inconsistent or implausible on its face that a reasonable factfinder would not credit it. Where such factors are present, the court of appeals may well find clear error_”).
Lieber is the only new facility at which double-celling has yet occurred. From the time double-celling began in February 1987 through July 1987, there were seven assaults without weapons and five with weapons.6 In contrast, from its opening in June 1986 through January 1987, when inmates were single-celled, there were twelve assaults without weapons and five with weapons. And while Commissioner Catoe stated generally that increased population complicates management, he assured the court that the State would be able to provide the programs and opportunities required by the decree if Lieber were double-celled.
V.
While prison overcrowding has been an increasing problem nationwide, remedial measures are now being taken under numerous court orders and consent decrees. However, correctional authorities attempting to comply with these orders and decrees are faced with the prevalent problem of increased prison populations. As a result the courts are, with increasing frequency, addressing requests for modification. The modification granted here is consistent with decisions rendered by this court and others under similar requests.
In Nelson v. Collins we reversed the district court’s refusal to approve double-celling at a modern, new facility in order to *215alleviate overcrowding at older facilities. 659 F.2d at 429. There, the Maryland State Prison System was directed by court order to relieve unconstitutional double-celling at two facilities and had entered into a consent decree regarding another facility. Subsequent to these decrees, the Supreme Court issued its decisions in Bell v. Wolfish, 441 U.S. 520, 99 S.Ct. 1861, 60 L.Ed.2d 447 (1979), and Rhodes v. Chapman, 452 U.S. 337, 101 S.Ct. 2392, 69 L.Ed.2d 59 (1981), clarifying the requirements of the eighth amendment as applied to correctional facilities. In addition, there were substantial changes in the conditions in the Maryland prison system. As here, Maryland had begun new construction in a concerted, good faith effort to meet the deadlines imposed by the decrees. But at the same time, Maryland also was faced with an unanticipated increase in its prison population. Taking into consideration the changed circumstances, this court found that the state was entitled to a modification in view of its good faith efforts and the clearly constitutional standards maintained at the new facility. 659 F.2d at 429.
Similarly, in Duran v. Elrod, the Seventh Circuit granted modification of a consent decree to allow double-celling in a county jail, reversing the district court’s refusal to do so. 760 F.2d at 763. Although the county had been dilatory in renovation and expansion of the jail, the court still found the modification reasonable in light of an “unremitting” increase in jail population and the potential harm to the citizens of the county. Id. at 761-62.
In contrast, modification of a consent decree regarding overcrowding in Texas prisons was refused in Ruiz v. Lynaugh, 811 F.2d 856 (5th Cir.1987). The court found that a foreseeable and ordinary increase in prison population did not justify modification. Id. at 862. Further, the requested modification would have allowed housing of inmates at facilities which did not meet other standards for basic services and conditions. See id. at n. 10. The conditions existing in the Texas prison system which supported the Fifth Circuit’s refusal to allow modification are not present in the case before us. Unlike Texas, the increase in prison population was unanticipated. In addition, except for double-celling at five new facilities, all terms and conditions of the consent decree as well as all constitutional standards are being met by the State of South Carolina.
VI.
The State of South Carolina voluntarily entered into a consent decree agreeing to certain terms and obligations and there is an initial inclination to hold the State to the letter of their agreement, as the district court did. However, this is not an ordinary contract between private parties to be strictly enforced. It is a compromise between the State and its inmates aimed at providing constitutionally adequate prison conditions. And, a flexible approach must be taken in addressing requests for modification.
The Department of Corrections and the General Assembly have demonstrated considerable good faith in attempting to attain total compliance. But, they have faced an unanticipated increase in inmate population, a problem common to prison systems across the nation. For example, the state-of-the-art prototype federal facility at Phoenix, on which the State institutions are modeled, was intended to provide single-celling, but within the first year of its opening, all inmates were double-celled. And despite the increases in population, the State has achieved the central goal of the decree — constitutionally adequate conditions of confinement.
Further, any solution to the problem not only affects the inmates and the Department of Corrections, but also the general public. Strict compliance would impose on the citizens of South Carolina a system which would increase the risk of danger to the general public if inmates who had not completed lawfully imposed sentences were prematurely discharged. Again, this risk far outweighs any imposition on inmates as a result of double-celling in the five new facilities described above.
The order of compliance is vacated and the case is remanded to the district court *216for modification of the consent decree to allow double-celling at the five new facilities. The district court will retain jurisdiction with the authority to provide appropriate remedies should the double-celling result in unconstitutional conditions of confinement.
VACATED and REMANDED with instructions.
. The five new facilities also contain segregation cells which house only one inmate and for which no modification is requested.
. Contrary to the assertion of the dissent, the State did not “baldly violate" the decree by constructing double cells smaller than the decree required. Infra at 218. It is undisputed that the cells were intended as single cells. But, it has never been satisfactorily explained to the court why, in the face of an agreement which required single cells to contain 50 square feet and double cells to contain 100 square feet, the Department of Corrections approved plans for single cells to contain 69 or 73 square feet.
. Judge Phillips argued against such an “overly rigid standard for relief’ in his dissent in Holiday Inns, urging a flexible standard "greatly dependent upon the particular facts of the case." *212645 F.2d at 244, 245. He read Swift & Co. "simply as a description of the proper exercise of those powers in the particular context of the case before that Court rather than as a general prescription for their exercise in all cases." Id. at 245. We agree that the general teaching of Swift & Co. "is merely that harm and continuing need must always be weighed in the balance in deciding whether continued enforcement of any injunctive decree is equitable in the light of specific changed circumstances." Id. As Judge Phillips further noted, and as evidenced in Nelson v. Collins, this court has “been properly selective and flexible in our application of the Swift & Co. standard, depending upon the nature of the litigation, the injunction at issue and the scope of the relief sought." Id.
. This is shown by statistics compiled by the South Carolina Department of Parole and Community Corrections.
. This failure rate refers to inmates returned to the custody of the Department of Corrections for convictions on new offenses with sentences of more than 90 days. This rate does not in-*214elude revocations for technical parole violations.
. These incidents reported from February 1987 through July 1987 represent only those assaults committed by general population inmates who are the only ones double-celled at Lieber.