DIANA GRIBBON MOTZ, Circuit Judge:

The district court refused to terminate a consent order, which the parties entered into in 1992, which became fully effective in 1994, and which provided that the court would retain jurisdiction over its subject matter until 1998. We affirm.

I.

This case is a class action by applicants for Aid to Families with Dependent Children (AFDC) and Medical Assistance (Medicaid) against state officials responsible for the administration of these programs in North Carolina and their agents, the administrators of the one hundred North Carolina county departments of social services. Since the inception of this case in 1974, the applicants have alleged, and the district court has repeatedly found, that the administrators have failed to comply with federal regulations concerning the processing of aid applications.

The administrators have been subject to numerous court orders and settlement agreements, all designed to encourage them to comply with federal law. The present conflict relates solely to a 1992 consent order, the terms of which the parties negotiated for many months. The stated purpose of the consent order was to bring all local social service departments "into compliance with the requirement in federal law to timely process AFDC and Medicaid applications without improper discouragement, denial or withdrawal of those applications." Pursuant to that purpose, the administrators agreed, inter alia, to meet the deadlines federal regulations mandate for processing applications, see 42 C.F.R. Section(s) 435.911; 45 C.F.R. Section(s) 206.10, and to monitor local departments' compliance. To calculate compliance, the parties agreed to substitute "hard" numbers where federal regulations were less quantifiable. The requirements of the consent order were phased in beginning August, 1992 and took full effect on January 1, 1994. The consent order also included a "sunset provision," in which the parties agreed that the district court would retain jurisdiction "for a period of six years from entry of this order," i.e., from August 1, 1992 until August 1, 1998.

In August, 1994, the administrators filed a motion to modify the consent order pursuant to Rule 60(b) of the Federal Rules of Civil Procedure. The administrators asserted that "unforeseen factual conditions resulting in unanticipated consequences have made implementation and compliance with the . . . consent order substantially more onerous and unworkable . . . ." The administrators maintained that the standard the Supreme Court employed in Rufo v. Inmates of Suffolk County Jail, 502 U.S. 367 (1992), applied to their motion and entitled them to a modification of the consent order. In the administrators' view, the Rufo standard was the "proper standard for modification or termination of consent decrees under Rule 60(b)." (emphasis added). The applicants opposed the motion, maintaining that the administrators could not meet the Rufo standard.

In March, 1995, without replying to the applicants' opposition or obtaining a ruling on the modification motion, the administrators filed a second Rule 60(b) motion, this time requesting that the district court terminate the consent order. Seven months after asserting that compliance with the consent order was impossible because its terms were "onerous" and "unworkable," the administrators now claimed that they had "complied in good faith" with the consent order. Rather than the Rufo standard they had previously urged on the court, the administrators asserted that the proper standard for evaluating their termination motion was set forth in Board of Education of Oklahoma City Public Schools v. Dowell, 498 U.S. 237 (1991). Under the Dowell standard, the administrators maintained that their good faith compliance with the consent order warranted its termination. They further informed the court that they had devised a plan to"ensure timely and efficient service to AFDC and Medicaid applicants," and represented that they would implement the plan if the court terminated the consent order.

The district court denied the administrators' motion to terminate. The court noted the parties' specific agreement in the 1992 consent order that the court would retain jurisdiction over the parties and "the subject matter of this action for a period of six years." Citing Rufo, the court found that the administrators were not entitled to relief from the order because they had failed to establish the existence of a "significant change in factual conditions or in the law" or to demonstrate that their proposed alternative plan was "suitably tailored" to any changed circumstance.

On appeal, the administrators argue that the district court erred in applying what they characterize as the "more stringent" Rufo standard, rather than the Dowell standard; they maintain that had the court properly applied the Dowell standard, they would have been entitled to termination of the consent order. These arguments exhibit both a fundamental misunderstanding of Dowell and Rufo and an effort to disregard the undisputed facts of the case at hand.

II.

The standards employed in Dowell and Rufo are but variations on a single theme. Both are grounded in the established general equity powers of the federal courts. Those powers, now formalized in Rule 60(b), which provided the basis for the motions made in Dowell, Rufo, and the case at hand, permit courts to grant parties relief from final judgments that have prospective effect. Rule 60(b) states, in pertinent part, that on "such terms as are just" a court may "relieve a party" from a judgment or order "where it is no longer equitable" that the judgment have "prospective application." Fed.R.Civ.P. 60(b)(5), (6). Accordingly, when confronted with any motion invoking this rule, a district court's task is to determine whether it remains equitable for the judgment at issue to apply prospectively and, if not, to relieve the parties of some or all of the burdens of that judgment on "such terms as are just."

The administrators seek to place Dowell and Rufo in separate compartments, with Dowell setting forth the standard to be applied when considering any and every motion to terminate a decree, and Rufo setting forth a "more stringent" standard for any and every motion to modify. However, a holding that motions to modify always require satisfaction of a more stringent standard than motions to terminate would be illogical. To adopt the administrators' argument would mean that the Supreme Court has mandated that the standard a party must meet to obtain the less serious remedy — modification of a permanent injunction — in every case is more rigorous than that needed to obtain the more drastic remedy — termination of such an injunction. In fact, examination of Dowell and Rufo makes clear that while the Rufo standard differs from the Dowell standard, it is hardly "more stringent." Analysis of these cases demonstrates that the Court articulated different standards not, as the administrators suggest, in response to the differences in the remedies requested (termination versus modification) but in recognition of the differences in the character and purposes of the injunctions at issue in the two cases. The administrators' myopic focus on the remedy requested ignores the genesis of Dowell and Rufo and the flexibility that they teach.

Prior to Dowell and Rufo, federal courts generally looked to United States v. Swift Co., 286 U.S. 106 (1932), for the standard to apply when reviewing motions to terminate or modify permanent injunctions. Swift involved an antitrust consent decree, in which meatpacking companies agreed to an injunction against certain monopolistic practices. Ten years later, the companies sought a modification of the decree, which the district court granted in part. The Supreme Court reversed, explaining: "Nothing less than a clear showing of grievous wrong evoked by new and unforeseen conditions should lead us to change what was decreed after years of litigation with the consent of all concerned." Id. at 119. Some courts, including this one, recognized that Swift permitted a more flexible approach when a party sought relief from decrees "directed to events to come" rather than at "fully accrued" rights. See Nelson v. Collins, 659 F.2d 420, 424 (4th Cir. 1981) (en banc) (quoting Swift, 286 U.S. at 114). However, for the most part, federal courts applied the Swift "grievous wrong" standard rigidly, leading them almost invariably to deny motions to terminate or modify permanent injunctions.

In its 1991 Dowell decision, the Supreme Court held that the Swift "grievous wrong" standard was not "the proper standard to apply to injunctions entered in school desegregation cases." Dowell, 498 U.S. at 248. The Court reasoned that because school desegregation decrees were intended as "temporary measure[s] to remedy past discrimination" and were "not intended to operate in perpetuity," a more flexible approach was warranted. Id. at 247. The Court distinguished Swift as a case involving a "continuing danger" that the party asking for the modification would revert to its prior illegal conduct. Id. (citing United States v. United Shoe Mach. Corp., 391 U.S. 244, 248 (1968)). Because the court-imposed decree in Dowell was aimed primarily at remedying past unconstitutional conduct, and no evidence indicated that the local authority would revert to its former illegal behavior, the Court held that to obtain relief the local authority need not demonstrate the existence of a "grievous wrong" caused by "new and unforeseen conditions." Dowell, 498 U.S. at 247. In the context of deciding whether to "modify or dissolve a desegregation decree," the Supreme Court directed that Rule 60(b) relief should be granted if the local authority could establish that (1) it had "complied in good faith with the desegregation decree" (2) its compliance had lasted for "a reasonable period of time" and (3) "the vestiges of past discrimination had been eliminated to the extent practicable." Id. at 249-50. The very next term, the Supreme Court, noting that it was sounding "the same theme" as in Dowell, again rejected the Swift "grievous wrong" test in favor of a more "flexible" approach. Rufo, 502 U.S. at 380. Rufo involved a consent decree between prisoners and local prison authorities. The authorities moved to modify the decree because they believed a change in the factual circumstances had made compliance much more difficult. The lower courts refused to modify the decree; the Supreme Court reversed. The Court noted initially that the "ability of a district court to modify a decree in response to changed circumstances" was particularly important in "institutional reform litigation." Id. The Court emphasized, however, that "a party may obtain relief from a court order when `it is no longer equitable that the judgment should have prospective application,' not when it is no longer convenient to live with the terms of a consent decree." Id. at 383 (quoting Fed.R.Civ. 60(b)(5)). Thus, even though a more flexible approach than Swift was warranted, in order to obtain the modification a party must demonstrate that (1) a"significant change" in the law or facts "warrants revision of the decree" and (2) the proposed modification is suitably tailored to the changed circumstances. Id.

Although in Dowell and Rufo the Court set forth different standards, its approach was the same. In both cases, the Court eschewed Swift's rigid "grievous wrong" standard in favor of a more flexible approach appropriate to the situation. In both, the Court analyzed a number of factors to determine whether the movants were entitled to Rule 60(b) relief. In both, the Court focused not on the remedy requested but on the changed equities of the situation and the nature of the injunctions involved.

Dowell involved a somewhat unusual situation: the district court had imposed an injunction not to ensure current compliance with federal law but primarily to remedy the effects of past wrongdoing (state-sponsored segregation). It was in this context that the Court held that Rule 60(b) relief would be justified if a party could establish that it had complied with the injunction for a reasonable period of time, so the effects of past wrongdoing were eliminated to the extent practicable, and therefore the injunction had accomplished its purpose. Rufo, on the other hand, involved not a judicially imposed injunction but a consent decree concerning the parties' agreement to cease current wrongdoing (prison over-crowding). In seeking Rule 60(b) relief, the prison officials in Rufo did not claim that they had complied with the consent decree or that it had accomplished its purpose. Rather, they maintained, and the Court held, that a significant change in circumstances would justify equitable relief. Thus, in Dowell and Rufo the Court fashioned standards triggered not by the precise remedy requested, but by the equitable showing made by the parties.

Nothing in Dowell or Rufo indicates that the Supreme Court intended the Rufo standard to apply to all modification motions and the Dowell standard to all termination motions. But see United States v. City of Miami, 2 F.3d 1497, 1503-05, 1508-09 (11th Cir. 1993) (directing district court to apply Rufo standard to modification motion and Dowell standard to termination motion); Heath v. DeCourcy, 992 F.2d 630, 633-35 (6th Cir. 1993) (same). Indeed, Dowell itself makes clear that a party can invoke the standard it sets forth to terminate or modify a decree. See Dowell, 498 U.S. at 248 (rejecting lower court's position that compliance alone "cannot become the basis for modifying or dissolving an injunction") (emphasis added); see also id. at 249 ("in deciding whether to modify or dissolve a desegregation decree, . . . compliance with previous court orders is obviously relevant") (emphasis added). Similarly, if a party can demonstrate a significant, unforeseen change in the facts or law, it may invoke Rufo to move for modification and even, if the unforeseen change in circumstances is particularly dramatic, termination. See, e.g., Sweeton v. Brown, 27 F.3d 1162, 1166 (6th Cir. 1994) (en banc), cert. denied, 115 S.Ct. 1118 (1995); Evans v. City of Chicago, 10 F.3d 474, 483 (7th Cir. 1993) (en banc) (Ripple, J., concurring), cert. denied, 114 S.Ct. 1831 (1994).

Thus, it is clear that Dowell and Rufo are entirely consistent; they do, indeed, sound the "same theme." Rufo, 502 U.S. at 380.

III.

What is unclear is whether the Dowell standard can be invoked when a party seeks to terminate or modify a consent decree, like that at issue here, which contains a specific sunset provision and was designed to remedy ongoing wrongdoing and to ensure present and future compliance with federal law.

A critical factor in the 1992 consent order, duly emphasized by the district court, is the administrators' agreement to abide by the terms of the order for six years. Unlike the injunction in Dowell, the order in this case was not court-imposed but resulted from extended negotiations between the parties. Of course, such an order is "enforceable as a judicial decree" and is therefore subject to Rule 60(b) like other judgments and decrees. Rufo, 502 U.S. at 378. However, "a consent decree embodies an agreement of the parties and thus in some respects is contractual in nature." Id.; see also United States v. ITT Cont. Baking Co., 420 U.S. 223, 236-37 (1975). By consenting to the terms of a decree, parties forego litigation and compromise their respective positions. For example, in this case, if the administrators had refused to enter into the consent decree, they might have avoided a judgment in which the district court required them to comply with the terms of the decree for six years; however, by agreeing to the consent decree with the six-year provision, the administrators avoided the possibility of receiving a court-imposed injunction that could have been more onerous in this or other respects.

Thus, defendants have somewhat different obligations under consent decrees than they do under court-imposed injunctions. In the latter, defendants can only be required to address ongoing illegal activity or the past effects of illegal activity — i.e., violations of substantive federal law. But, in a consent decree, defendants may agree, within limits, to do more than a judicially imposed injunction could have required. See Plyler v. Evatt, 924 F.2d 1321, 1327 (4th Cir. 1991). The consent decree must, of course, still relate to an alleged violation of federal law, see Rufo, 502 U.S. at 389, but if it does, the parties should be held to their bargain. See id. at 391-92 (A "court should not `turn aside to inquire whether some of [the provisions of the decree] upon separate as distinguished from joint action could have been opposed with success if the defendants had offered opposition.'") (quoting Swift, 286 U.S. at 116-17). It would seem that the six-year term makes it impossible for the administrators to have complied with the 1992 consent order, or for the order to have fulfilled its purpose, in less than six years.

Furthermore, like the consent decree in Rufo, see 502 U.S. at 373, the consent order at issue here was designed to remedy ongoing wrongdoing and to ensure present and future compliance with federal law. In contrast, as noted previously, the injunction in Dowell was fashioned primarily to remedy the effects of past illegal conduct. See Dowell, 498 U.S. at 240 (busing ordered to remedy past segregation). When a decree's purpose is to ensure ongoing and future compliance with federal law, it is obviously much more difficult to establish that the decree has fulfilled that purpose (especially after the decree has been in effect a relatively short time) than when its principal purpose is to remedy the effects of past illegal activity.

Other circuits have recognized that the Dowell standard does not appear suitable for decrees aimed at ensuring ongoing compliance with the law. Thus, the Dowell standard has been applied primarily in cases, like Dowell itself, involving decrees that may generally enjoin ongoing illegal activity, but are directed principally at remedying the effects of past illegal activity. See, e.g., Youngblood v. Dalzell, 925 F.2d 954, 955-56 (6th Cir. 1991). No court has employed Dowell to terminate or modify a decree aimed exclusively, or even principally, at ongoing illegal activity. In those instances it is to Rufo, rather than Dowell, that our sister circuits have turned not only to determine whether such decrees should be modified but also whether they should be terminated. See, e.g., United States v. Eastman Kodak Co., 63 F.3d 95, 102 (2d Cir. 1995) (relying on Rufo and United Shoe); Sweeton v. Brown, 27 F.3d 1162, 1163-64 (6th Cir. 1994) (en banc), cert. denied, 115 S.Ct. 1118 (1995); Protectoseal Co. v. Barancik, 23 F.3d 1184, 1187 (7th Cir. 1994); Evans v. City of Chicago, 10 F.3d 474, 476 (7th Cir. 1993) (en banc).

Accordingly, the Dowell standard may well be inapplicable to consent decrees, like the 1992 consent order, in which the parties agree to cease ongoing illegal activity under specified terms and conditions. We note that the First Circuit has voiced similar uncertainty as to the applicability of Dowell in this context. See Inmates of Suffolk County Jail v. Rufo, 12 F.3d 286, 292-93 (1st Cir. 1993) (after remand from the Supreme Court). The uncertainty would appear to be exacerbated here where the consent decree specifically provides for court supervision for a defined, and relatively short, period of time. We need not definitively resolve the question in this case because in any event, even if the Dowell standard applied to the 1992 consent order, the administrators did not and could not satisfy that standard.

IV.

As discussed above, to meet the Dowell standard, the administrators must demonstrate: (1) that for a reasonable period of time (2) they have complied in good faith with the consent decree (3) to the point that the "vestiges" of past unlawful behavior have been eliminated "to the extent practicable," and thus the purpose of the decree has been satisfied. See Dowell, 498 U.S. at 249-50.

The administrators studiously ignore the first of these factors. The reason seems clear: the record unequivocally demonstrates that the administrators have not complied with the consent decree for a reasonable period of time. The 1992 consent decree was issued less than four years ago and has only been fully effective since January, 1994. Thus, in March, 1995, when the administrators moved to terminate the order, it had been in full effect for little more than a year. No court has held that compliance for such a short period constitutes compliance for a "reasonable period" of time.

Only compliance for substantially longer periods has been regarded as significant evidence of good faith compliance. See Dowell, 498 U.S. at 249 (the "passage of time [twelve years] enables the District Court to observe the good faith of the school board in complying with the decree"); see also Patterson v. Newspaper Mail Deliverers' Union, 13 F.3d 33, 34-35 (2d Cir. 1993) (motion to terminate, denied eleven years after a decree's entry, was properly granted after eighteen years), cert. denied, 115 S.Ct. 58 (1994); City of Miami, 2 F.3d at 1508 (after consent decree had been in effect for fifteen years, district court should examine whether termination warranted); Consumer Advisory Bd. v. Glover, 989 F.2d 65, 68 (1st Cir. 1993) (after fifteen years, district court held to have "considerable discretion" to conclude the decree should be dissolved). Cf. Youngblood, 925 F.2d at 955 (although seventeen years had passed, district court should reconsider its decision to terminate the decree).

Moreover, even if the parties' agreement to the six-year sunset provision does not mandate that the consent order remain effective for that entire period, see Collins v. Thompson, 8 F.3d 657, 659 (9th Cir. 1993), cert. denied, 114 S.Ct. 2133 (1994); In re Pearson, 990 F.2d 653, 658 (1st Cir. 1993); Glover, 989 F.2d at 67, it does obviously merit significant weight in a court's consideration of whether the order has been in effect for a "reasonable period." This is because the parties themselves typically will be the most knowledgeable as to the reasonable length of time necessary to determine whether the decree has had its desired effect. Furthermore, ignoring the parties' agreement in a consent order as to the appropriate time period would reduce (if not destroy) the incentive for parties to enter into consent decrees in the first place. See Patterson, 13 F.3d at 38 (although flexible standard applies to modification and termination motions, "it is also important to provide all concerned with an incentive to enter into constructive settlements so that protracted litigation can be avoided and useful remedies developed by agreement, rather than by judicial command"); see also Rufo, 502 U.S. at 389-90 (noting importance of finality as an incentive for parties to negotiate settlements).

Medicaid and AFDC applicants have been forced to litigate for more than twenty years in an effort to make the administrators comply with federal law. The administrators themselves concede that they have had significant difficulty in adhering to federal regulations. They acknowledge that "[c]ertainly [judicial] supervision may have been justified throughout part of the 1970's, the 1980's and even into the 1990's." Brief of Appellants at 3. Thus, in this case, a party that has had difficulty complying with federal law for an extended period consented in writing to abide for six years with the terms of a decree aimed at ensuring ongoing compliance. Under such circumstances, it is not possible for such a party to establish at the conclusion of less than two years that it has complied with the decree for a "reasonable period of time" sufficient to establish that the decree has served its purpose.

This brings us to a second factor the Supreme Court emphasized in Dowell — a party's good faith compliance with the decree. Although we express no opinion as to whether the administrators have acted in good faith, the evidence is uncontroverted that the administrators have never been able to comply with the consent order. Indeed, the administrators' own statistics document their failure to achieve compliance. The central purpose and requirement of the consent order was for "all applications [to] be processed timely in compliance with federal law." To that end, the consent order requires each county department to meet both an "average processing time" and a "percent processed timely" threshold each month. As the administrators' statistics reveal, since the inception of the order there has never been a month in which all one hundred county departments complied with the thresholds agreed to in the order. On average, since the time the order became fully effective in January, 1994, one-fifth of the local departments have failed to comply with these thresholds. In the two most recent months for which the administrators submitted data, January, 1995 and February, 1995, thirty-six counties and twenty counties, respectively, failed to comply with the thresholds.

The administrators respond that the statistics must be analyzed in context, arguing that "most of the departments out of compliance were in small, less populated counties," and so it is unfair to focus on them. Brief of Appellants at 4-5. Whatever the merits of this claim, the fact that the administrators need to make it at all reveals, in the administrators' own words, that they have been continually "out of compliance" with the consent order.

The administrators seem to believe that the Dowell standard permits good intentions to substitute for compliance. Thus, before us they claim, not that they have "complied with the consent order," but that they have "complied in good faith with the basic purpose of the litigation as embodied in the 1992 Consent Order." Brief of Appellants at 17. The administrators misread Dowell to hold that when a party complies with a consent order "to the extent practicable," the party is entitled to Rule 60(b) relief. Dowell requires more. A party seeking to terminate a decree must demonstrate "compliance with it." Dowell, 498 U.S. at 248-51. See also Heath, 992 F.2d at 634 ("full compliance" must be ensured by the district court prior to terminating an injunction). Good faith is not a substitute for compliance; rather, compliance demonstrates good faith. See Dowell, 498 U.S. at 249 (the district court can "observe the good faith of the school board in complying with the decree."). Without proof of a reasonable period of compliance, regardless of a party's subjective good faith or good intentions, equitable considerations weigh strongly against terminating a consent order.

The last factor in the Dowell standard is whether the administrators have removed the vestiges of past wrongdoing to the extent that the purpose of the consent order has been fulfilled. For the same reasons that it is unclear whether the Dowell standard applies at all to the 1992 consent order, the administrators cannot make this showing. The primary purpose of the order in this case was not to redress past wrongdoing but to ensure current and future compliance with federal regulations for the period between 1992-1998. Due to the inherent difficulty in proving that the purpose of a decree ensuring ongoing compliance for six years has been fulfilled prior to the end of the six year period, especially when the administrators cannot establish that they have ever fully complied with the consent order, the administrators cannot satisfy this element either.

V.

In view of the fact that the administrators can establish none of the Dowell factors, the Dowell standard, even if applicable here, is of no use to them. Because the administrators could not meet the Dowell standard, the district court correctly analyzed their termination motion under the Rufo standard. Indeed, under the circumstances of this case, the application of Rufo provides the flexible approach to Rule 60(b) that the Supreme Court has urged. Particularly, in view of the litigation history of this case, there was no error in applying Rufo.

It will be remembered that only seven months prior to moving to terminate the decree, the administrators had filed a motion to modify, which remained pending when the court resolved the termination motion. In the modification motion, the administrators acknowledged that Rufo set forth the standard for modification or termination of consent decrees." Moreover, attached to the termination motion was a new plan, devised by the administrators, for processing aid applications. Although the administrators did not expressly urge the court to modify the consent order by substituting in its place the voluntary plan, they certainly did assert that the voluntary plan would effectively substitute for the consent order.

The administrators thus have been less than clear both as to whether they were invoking Dowell or Rufo and whether they were pursuing elimination of all court-approved plans or substitution of a new plan. In view of this and the total lack of any support in the record that the Dowell standard (even if applicable) had been met, the district court's examination of the administrators' motion under the Rufo standard can hardly be considered error. Rather, this appears to be precisely the sort of flexibility appropriate to an equitable analysis of a Rule 60(b) motion.

Applying Rufo, the district court found that the administrators had failed to establish a significant change in the law or facts that warranted relief from the 1992 consent order. The administrators have not challenged the court's finding under Rufo and, therefore, that issue is not before us. Nevertheless, we reiterate the district court's observation that should Congress pass legislation relating to the programs at issue in this case, the administrators are free, of course, to re-file motions for relief under Rule 60(b). Likewise, should the administrators be able to establish a significant change in the factual circumstances, relief under Rule 60(b) could be reassessed.

VI.

For all of these reasons, the district court's order denying the motion to terminate is AFFIRMED.