delivered the opinion of the Court.
DeKalb County, Georgia, is a major suburban area of Atlanta. This case involves a court-ordered desegregation decree for the DeKalb County School System (DCSS). DCSS now serves some 73,000 students in kindergarten through high school and is the 32d largest elementary and secondary school system in the Nation.
DCSS has been subject to the supervision and jurisdiction of the United States District Court for the Northern District of Georgia since 1969, when it was ordered to dismantle its dual school system. In 1986, petitioners filed a motion for final dismissal. The District Court ruled that DCSS had not achieved unitary status in all respects but had done so in student attendance and three other categories. In its order the District Court relinquished remedial control as to those aspects of the system in which unitary status had been achieved, and retained supervisory authority only for those aspects of the school system in which the district was not in full compliance. The Court of Appeals for the Eleventh Circuit reversed, 887 F. 2d 1438 (1989), holding that a district court should retain full remedial authority over a school system until it achieves unitary status in six categories at the same time for several years. We now reverse the judgment of the Court of Appeals and remand, holding that a district court is permitted to withdraw judicial supervision with respect to discrete categories in which the school district has achieved compliance with a court-ordered desegregation plan. A district court need not retain active control over every aspect of school administration until a school district has demonstrated unitary status in all facets of its system.
HH
<C
For decades before our decision in Brown v. Board of Education. 347 U. S. 483 (1954) (Brown I), and our mandate in *472Brown v. Board of Education, 349 U. S. 294, 301 (1955) (Brown II), which ordered school districts to desegregate with “all deliberate speed,” DCSS was segregated by law. DCSS’ initial response to the mandate of Brown II was an all too familiar one. Interpreting “all deliberate speed” as giving latitude to delay steps to desegregate, DCSS took no positive action toward desegregation until the 1966-1967 school year, when it did nothing more than adopt a freedom of choice transfer plan. Some black students chose to attend former de jure white schools, but the plan had no significant effect on the former de jure black schools.
In 1968, we decided Green v. School Bd. of New Kent County, 391 U. S. 430. We held that adoption of a freedom of choice plan does not, by itself, satisfy a school district’s mandatory responsibility to eliminate all vestiges of a dual system. Green was a turning point in our law in a further respect. Concerned by more than a decade of inaction, we stated that “ ‘[t]he time for mere “deliberate speed” has run out.’ ” Id., at 438, quoting Griffin v. Prince Edward County School Bd., 377 U. S. 218, 234 (1964). We said that the obligation of school districts once segregated by law was to come forward with a plan that “promises realistically to work, and promises realistically to work now.” 391 U. S., at 439 (emphasis in original). The case before us requires an understanding and assessment of how DCSS responded to the directives set forth in Green.
Within two months of our ruling in Green, respondents, who are black schoolchildren and their parents, instituted this class action in the United States District Court for the Northern District of Georgia. After the suit was filed, DCSS voluntarily began working with the Department of Health, Education, and Welfare to devise a comprehensive and final plan of desegregation. The District Court, in June 1969, entered a consent order approving the proposed plan, which was to be implemented in the 1969-1970 school year. The order abolished the freedom of choice plan and adopted *473a neighborhood school attendance plan that had been proposed by DCSS and accepted by the Department of Health, Education, and Welfare subject to a minor modification. Under the plan all of the former de jure black schools were closed, and their students were reassigned among the remaining neighborhood schools. The District Court retained jurisdiction.
Between 1969 and 1986, respondents sought only infrequent and limited judicial intervention into the affairs of DCSS. They did not request significant changes in student attendance zones or student assignment policies. In 1976, DCSS was ordered to expand its Majority-to-Minority (M-to-M) student transfer program, allowing students in a school where they are in the majority race to transfer to a school where they are in the minority; to establish a biracial committee to oversee the transfer program and future boundary line changes; and to reassign teachers so that the ratio of black to white teachers in each school would be, in substance, similar to the racial balance in the school population systemwide. From 1977 to 1979, the District Court approved a boundary line change for one elementary school attendance zone and rejected DCSS proposals to restrict the M-to-M transfer program. In 1983, DCSS was ordered to make further adjustments to the M-to-M transfer program.
In 1986, petitioners filed a motion for final dismissal of the litigation. They sought a declaration that DCSS had satisfied its duty to eliminate the dual education system, that is to say a declaration that the school system had achieved unitary status. Green, supra, at 441. The District Court approached the question whether DCSS had achieved unitary status by asking whether DCSS was unitary with respect to each of the factors identified in Green. The court considered an additional factor that is not named in Green: the quality of education being offered to the white and black student populations.
*474The District Court found DCSS to be “an innovative school system that has travelled the often long road to unitary status almost to its end,” noting that “the court has continually been impressed by the successes of the DCSS and its dedication to providing a quality education for all students within that system.” App. to Pet. for Cert. 71a. It found that DCSS is a unitary system with regard to student assignments, transportation, physical facilities, and extracurricular activities, and ruled that it would order no further relief in those areas. The District Court stopped short of dismissing the case, however, because it found that DCSS was not unitary in every respect. The court said that vestiges of the dual system remain in the areas of teacher and principal assignments, resource allocation, and quality of education. DCSS was ordered to take measures to address the remaining problems.
B
Proper resolution of any desegregation case turns on a careful assessment of its facts. Green, supra, at 439. Here, as in most cases where the issue is the degree of compliance with a school desegregation decree, a critical beginning point is the degree of racial imbalance in the school district, that is to say a comparison of the proportion of majority to minority students in individual schools with the proportions of the races in the district as a whole. This inquiry is fundamental, for under the former de jure regimes racial exclusion was both the means and the end of a policy motivated by disparagement of, or hostility towards, the disfavored race. In accord with this principle, the District Court began its analysis with an assessment of the current racial mix in the schools throughout DCSS and the explanation for the racial imbalance it found. Respondents did not contend on appeal that the findings of fact were clearly erroneous, and the Court of Appeals did not find them to be erroneous. The Court of Appeals did disagree with the conclusion reached *475by the District Court respecting the need for further supervision of racial balance in student assignments.
In the extensive record that comprises this case, one fact predominates: Remarkable changes in the racial composition of the county presented DCSS and the District Court with a student population in 1986 far different from the one they set out to integrate in 1969. Between 1950 and 1985, DeKalb County grew from 70,000 to 450,000 in total population, but most of the gross increase in student enrollment had occurred by 1969, the relevant starting date for our purposes. Although the public school population experienced only modest changes between 1969 and 1986 (remaining in the low 70,000’s), a striking change occurred in the racial proportions of the student population. The school system that the District Court ordered desegregated in 1969 had 5.6% black students; by 1986 the percentage of black students was 47%.
To compound the difficulty of working with these radical demographic changes, the northern and southern parts of the county experienced much different growth patterns. The District Court found that “[a]s the result of these demographic shifts, the population of the northern half of DeKalb County is now predominantly white and the southern half of DeKalb County is predominantly black.” App. to Pet. for Cert. 38a. In 1970, there were 7,615 nonwhites living in the northern part of DeKalb County and 11,508 nonwhites in the southern part of the county. By 1980, there were 15,365 nonwhites living in the northern part of the county, and 87,583 nonwhites in the southern part. Most of the growth in the nonwhite population in the southern portion of the county was due to the migration of black persons from the city of Atlanta. Between 1975 and 1980 alone, approximately 64,000 black citizens moved into southern DeKalb County, most of them coming from Atlanta. During the same period, approximately 37,000 white citizens moved out of southern DeKalb County to the surrounding counties.
*476The District Court made findings with respect to the number of nonwhite citizens in the northern and southern parts of the county for the years 1970 and 1980 without making parallel findings with respect to white citizens. Yet a clear picture does emerge. During the relevant period, the black population in the southern portion of the county experienced tremendous growth while the white population did not, and the white population in the northern part of the county experienced tremendous growth while the black population did not.
The demographic changes that occurred during the course of the desegregation order are an essential foundation for the District Court’s analysis of the current racial mix of DCSS. As the District Court observed, the demographic shifts have had “an immense effect on the racial compositions of the DeKalb County schools.” Ibid. From 1976 to 1986, enrollment in elementary schools declined overall by 15%, while black enrollment in elementary schools increased by 86%. During the same period, overall high school enrollment declined by 16%, while black enrollment in high schools increased by 119%. These effects were even more pronounced in the southern portion of DeKalb County.
Concerned with racial imbalance in the various schools of the district, respondents presented evidence that during the 1986-1987 school year DCSS had the following features: (1) 47% of the students attending DCSS were black; (2) 50% of the black students attended schools that were over 90% black; (3) 62% of all black students attended schools that had more than 20% more blacks than the system-wide average; (4) 27% of white students attended schools that were more than 90% white; (5) 59% of the white students attended schools that had more than 20% more whites than the system-wide average; (6) of the 22 DCSS high schools, five had student populations that were more than 90% black, while five other schools had student populations that were more than 80% white; and (7) of the 74 elementary schools *477in DCSS, 18 are over 90% black, while 10 are over 90% white. Id., at 31a. (Respondents’ evidence on these points treated all nonblack students as white. The District Court noted that there was no evidence that nonblack minority students constituted even 1% of DCSS student population.)
Respondents argued in the District Court that this racial imbalance in student assignment was a vestige of the dual system, rather than a product of independent demographic forces. In addition to the statistical evidence that the ratio of black students to white students in individual schools varied to a significant degree from the system-wide average, respondents contended that DCSS had not used all available desegregative tools in order to achieve racial balancing. Respondents pointed to the following alleged shortcomings in DCSS’ desegregative efforts: (1) DCSS did not break the county into subdistricts and racially balance each subdistrict; (2) DCSS failed to expend sufficient funds for minority learning opportunities; (3) DCSS did not establish community advisory organizations; (4) DCSS did not make full use of the freedom of choice plan; (5) DCSS did not cluster schools, that is, it did not create schools for separate grade levels which could be used to establish a feeder pattern; (6) DCSS did not institute its magnet school program as early as it might have; and (7) DCSS did not use busing to facilitate urban to suburban exchanges.
According to the District Court, respondents conceded that the 1969 order assigning all students to their neighborhood schools “effectively desegregated the DCSS for a period of time” with respect to student assignment. Id., at 36a. The District Court noted, however, that despite this concession respondents contended there was an improper imbalance in two schools even in 1969. Respondents made much of the fact that despite the small percentage of blacks in the county in 1969, there were then two schools that contained a majority of black students: Terry Mill Elementary School *478was 76% black, and Stoneview Elementary School was 51% black.
The District Court found the racial imbalance in these schools was not a vestige of the prior de jure system. It observed that both the Terry Mill and Stoneview schools were de jure white schools before the freedom of choice plan was put in place. It cited expert witness testimony that Terry Mill had become a majority black school as a result of demographic shifts unrelated to the actions of petitioners or their predecessors. In 1966, the overwhelming majority of students at Terry Mill were white. By 1967, due to migration of black citizens from Atlanta into DeKalb County — and into the neighborhood surrounding the Terry Mill school in particular — 23% of the students at Terry Mill were black. By 1968, black students constituted 50% of the school population at Terry Mill. By 1969, when the plan was put into effect, the percentage of black students had grown to 76. In accordance with the evidence of demographic shifts, and in the absence of any evidence to suggest that the former dual system contributed in any way to the rapid racial transformation of the Terry Mill student population, the District Court found that the pre-1969 unconstitutional acts of petitioners were not responsible for the high percentage of black students at the Terry Mill school in 1969. Its findings in this respect are illustrative of the problems DCSS and the District Court faced in integrating the whole district.
Although the District Court found that DCSS was desegregated for at least a short period under the court-ordered plan of 1969, it did not base its finding that DCSS had achieved unitary status with respect to student assignment on that circumstance alone. Recognizing that “[t]he achievement of unitary status in the area of student assignment cannot be hedged on the attainment of such status for a brief moment,” id., at 37a, the District Court examined the interaction between DCSS policy and demographic shifts in DeKalb County.
*479The District Court noted that DCSS had taken specific steps to combat the effects of demographics on the racial mix of the schools. Under the 1969 order, a biracial committee had reviewed all proposed changes in the boundary lines of school attendance zones. Since the original desegregation order, there had been about 170 such changes. It was found that only three had a partial segregative effect. An expert testified, and the District Court found, that even those changes had no significant effect on the racial mix of the school population, given the tremendous demographic shifts that were taking place at the same time.
The District Court also noted that DCSS, on its own initiative, started an M-to-M program in the 1972 school year. The program was a marked success. Participation increased with each passing year, so that in the 1986-1987 school year, 4,500 of the 72,000 students enrolled in DCSS participated. An expert testified that the impact of an M-to-M program goes beyond the number of students transferred because students at the receiving school also obtain integrated learning experiences. The District Court found that about 19% of the students attending DCSS had an integrated learning experience as a result of the M-to-M program. Id., at 40a.
In addition, in the 1980's, DCSS instituted a magnet school program in schools located in the middle of the county. The magnet school programs included a performing arts program, two science programs, and a foreign language program. There was testimony in the District Court that DCSS also had plans to operate additional magnet programs in occupational education and gifted and talented education, as well as a preschool program and an open campus. By locating these programs in the middle of the county, DCSS sought to attract black students from the southern part of the county and white students from the northern part.
Further, the District Court found that DCSS operates a number of experience programs integrated by race, including *480a writing center for fifth and seventh graders, a driving range, summer school programs, and a dialectical speech program. DCSS employs measures to control the racial mix in each of these special areas.
In determining whether DCSS has achieved unitary status with respect to student assignment, the District Court saw its task as one of deciding if petitioners “have accomplished maximum practical desegregation of the DCSS or if the DCSS must still do more to fulfill their affirmative constitutional duty.” Id., at 41a. Petitioners and respondents presented conflicting expert testimony about the potential effects that desegregative techniques not deployed might have had upon the racial mix of the schools. The District Court found that petitioners’ experts were more reliable, citing their greater familiarity with DCSS, their experience, and their standing within the expert community. The District Court made these findings:
“[The actions of DCSS] achieved maximum practical desegregation from 1969 to 1986. The rapid population shifts in DeKalb County were not caused by any action on the part of the DCSS. These demographic shifts were inevitable as the result of suburbanization, that is, work opportunities arising in DeKalb County as well as the City of Atlanta, which attracted blacks to DeKalb; the decline in the number of children born to white families during this period while the number of children born to black families did not decrease; blockbusting of formerly white neighborhoods leading to selling and buying of real estate in the DeKalb area on a highly dynamic basis; and the completion of Interstate 20, which made access from DeKalb County into the City of Atlanta much easier. . . . There is no evidence that the school system’s previous unconstitutional conduct may have contributed to this segregation. This court is convinced that any further actions taken by defendants, while the actions might have made marginal adjustments in the *481population trends, would not have offset the factors that were described above and the same racial segregation would have occurred at approximately the same speed.” Id., at 44a-45a.
The District Court added:
“[A]bsent massive bussing, which is not considered as a viable option by either the parties or this court, the magnet school program and the M-to-M program, which the defendants voluntarily implemented and to which the defendants obviously are dedicated, are the most effective ways to deal with the effects on student attendance of the residential segregation existing in DeKalb County at this time.” Id., at 46a.
Having found no constitutional violation with respect to student assignment, the District Court next considered the other Green factors, beginning with faculty and staff assignments. The District Court first found that DCSS had fulfilled its constitutional obligation with respect to hiring and retaining minority teachers and administrators. DCSS has taken active steps to recruit qualified black applicants and has hired them in significant numbers, employing a greater percentage of black teachers than the statewide average. The District Court also noted that DCSS has an “equally exemplary record” in retention of black teachers and administrators. App. to Pet. for Cert. 49a. Nevertheless, the District Court found that DCSS had not achieved or maintained a ratio of black to white teachers and administrators in each school to approximate the ratio of black to white teachers and administrators throughout the system. See Singleton v. Jackson Municipal Separate School Disk, 419 P. 2d 1211 (CA5 1969), cert. denied, 396 U. S. 1032 (1970). In other words, a racial imbalance existed in the assignment of minority teachers and administrators. The District Court found that in the 1984-1985 school year, seven schools deviated by more than 10% from the system-wide average *482of 26.4% minority teachers in elementary schools and 24.9% minority teachers in high schools. The District Court also found that black principals and administrators were overrepresented in schools with high percentages of black students and underrepresented in schools with low percentages of black students.
The District Court found the crux of the problem to be that DCSS has relied on the replacement process to attain a racial balance in teachers and other staff and has avoided using mandatory reassignment. DCSS gave as its reason for not using mandatory reassignment that the competition among local school districts is stiff, and that it is difficult to attract and keep qualified teachers if they are required to work far from their homes. In fact, because teachers prefer to work close to their homes, DCSS has a voluntary transfer program in which teachers who have taught at the same school for a period of three years may ask for a transfer. Because most teachers request to be transferred to schools near their homes, this program makes compliance with the objective of racial balance in faculty and staff more difficult.
The District Court stated that it was not “unsympathetic to the difficulties that DCSS faces in this regard,” but held that the law of the Circuit requires DCSS to comply with Singleton. App. to Pet. for Cert. 53a. The court ordered DCSS to devise a plan to achieve compliance with Singleton, noting that “[i]t would appear that such compliance will necessitate reassignment of both teachers and principals.” App. to Pet. for Cert. 58a. With respect to faculty, the District Court noted that meeting Singleton would not be difficult, citing petitioners’ own estimate that most schools’ faculty could conform by moving, at most, two or three teachers.
Addressing the more ineffable category of quality of education, the District Court rejected most of respondents’ contentions that there was racial disparity in the provision of certain educational resources (e. g., teachers with advanced *483degrees, teachers with more experience, library books), contentions made to show that black students were not being given equal educational opportunity. The District Court went further, however, and examined the evidence concerning achievement of black students in DCSS. It cited expert testimony praising the overall educational program in the district, as well as objective evidence of black achievement: Black students at DCSS made greater gains on the Iowa Tests of Basic Skills than white students, and black students at DCSS are more successful than black students nationwide on the Scholastic Aptitude Test. It made the following finding:
“While there will always be something more that the DCSS can do to improve the chances for black students to achieve academic success, the court cannot find, as plaintiffs urge, that the DCSS has been negligent in its duties to implement programs to assist black students. The DCSS is a very innovative school system. It has implemented a number of programs to enrich the lives and enhance the academic potential of all students, both blacks and whites. Many remedial programs are targeted in the majority black schools. Programs have been implemented to involve the parents and offset negative socio-economic factors. If the DCSS has failed in any way in this regard, it is not because the school system has been negligent in its duties.” App. to Pet. for Cert. 69a-70a (footnote omitted).
Despite its finding that there was no intentional violation, the District Court found that DCSS had not achieved unitary status with respect to quality of education because teachers in schools with disproportionately high percentages of white students tended to be better educated and have more experience than their counterparts in schools with disproportionately high percentages of black students, and because per-pupil expenditures in majority white schools *484exceeded per-pupil expenditures in majority black schools. From these findings, the District Court ordered DCSS to equalize spending and remedy the other problems.
The final Green factors considered by the District Court were: (1) physical facilities, (2) transportation, and (3) extracurricular activities. The District Court noted that although respondents expressed some concerns about the use of portable classrooms in schools in the southern portion of the county, they in effect conceded that DCSS has achieved unitary status with respect to physical facilities.
In accordance with its factfinding, the District Court held that it would order no further relief in the areas of student assignment, transportation, physical facilities, and extracurricular activities. The District Court, however, did order DCSS to establish a system to balance teacher and principal assignments and to equalize per-pupil expenditures throughout DCSS. Having found that blacks were represented on the school board and throughout DCSS administration, the District Court abolished the biracial committee as no longer necessary.
Both parties appealed to the United States Court of Appeals for the Eleventh Circuit. The Court of Appeals affirmed the District Court’s ultimate conclusion that DCSS has not yet achieved unitary status, but reversed the District Court’s ruling that DCSS has no further duties in the area of student assignment. 887 F. 2d 1438 (1989). The Court of Appeals held that the District Court erred by considering the six Green factors as separate categories. The Court of Appeals rejected the District Court’s incremental approach, an approach that has also been adopted by the Court of Appeals for the First Circuit, Morgan v. Nucci, 831 F. 2d 313, 318-319 (1987), and held that a school system achieves unitary status only after it has satisfied all six factors at the same time for several years. 887 F. 2d, at 1446. Because, under this test, DCSS had not achieved unitary status at any time, the Court of Appeals held that DCSS could “not shirk *485its constitutional duties by pointing to demographic shifts occurring prior to unitary status.” Id., at 1448. The Court of Appeals held that petitioners bore the responsibility for the racial imbalance, and in order to correct that imbalance would have to take actions that “may be administratively awkward, inconvenient, and even bizarre in some situations,” Swann v. Charlotte-Mecklenburg Bd. of Education, 402 U. S. 1, 28 (1971), such as pairing and clustering of schools, drastic gerrymandering of school zones, grade reorganization, and busing. We granted certiorari, 498 U. S. 1081 (1991).
II
Two principal questions are presented. The first is whether a district court may relinquish its supervision and control over those aspects of a school system in which there has been compliance with a desegregation decree if other aspects of the system remain in noncompliance. As we answer this question in the affirmative, the second question is whether the Court of Appeals erred in reversing the District Court’s order providing for incremental withdrawal of supervision in all the circumstances of this case.
A
The duty and responsibility of a school district once segregated by law is to take all steps necessary to eliminate the vestiges of the unconstitutional de jure system. This is required in order to ensure that the principal wrong of the de jure system, the injuries and stigma inflicted upon the race disfavored by the violation, is no longer present. This was the rationale and the objective of Brown I and Brown II. In Brown I we said: “To separate [black students] from others of similar age and qualifications solely because of their race generates a feeling of inferiority as to their status in the community that may affect their hearts and minds in a way unlikely ever to be undone.” 347 U. S., at 494. We *486quoted a finding of the three-judge District Court in the underlying Kansas case that bears repeating here:
“ ‘Segregation of white and colored children in public schools has a detrimental effect upon the colored children. The impact is greater when it has the sanction of the law; for the policy of separating the races is usually interpreted as denoting the inferiority of the negro group. A sense of inferiority affects the motivation of a child to learn. Segregation with the sanction of law, therefore, has a tendency to [retard] the educational and mental development of negro children and to deprive them of some of the benefits they would receive in a racial[ly] integrated school system.’” Ibid.
The objective of Brown I was made more specific by our holding in Green that the duty of a former de jure district is to “take whatever steps might be necessary to convert to a unitary system in which racial discrimination would be eliminated root and branch.” 391 U. S., at 437-438. We also identified various parts of the school system which, in addition to student attendance patterns, must be free from racial discrimination before the mandate of Brown is met: faculty, staff, transportation, extracurricular activities, and facilities. 391 U. S., at 435. The Green factors are a measure of the racial identifiability of schools in a system that is not in compliance with Brown, and we instructed the District Courts to fashion remedies that address all these components of elementary and secondary school systems.
The concept of unitariness has been a helpful one in defining the scope of the district courts’ authority, for it conveys the central idea that a school district that was once a dual system must be examined in all of its facets, both when a remedy is ordered and in the later phases of desegregation when the question is whether the district courts’ remedial control ought to be modified, lessened, or withdrawn. But, as we explained last Term in Board of Ed. of Oklahoma City *487Public Schools v. Dowell, 498 U. S. 237, 245-246 (1991), the term “unitary” is not a precise concept:
“[I]t is a mistake to treat words such as ‘dual’ and ‘unitary’ as if they were actually found in the Constitution. . . . Courts have used the terms ‘dual’ to denote a school system which has engaged in intentional segregation of students by race, and ‘unitary’ to describe a school system which has been brought into compliance with the command of the Constitution. We are not sure how useful it is to define these terms more precisely, or to create subclasses within them.”
It follows that we must be cautious not to attribute to the term a utility it does not have. The term “unitary” does not confine the discretion and authority of the District Court in a way that departs from traditional equitable principles.
That the term “unitary” does not have fixed meaning or content is not inconsistent with the principles that control the exercise of equitable power. The essence of a court’s equity power lies in its inherent capacity to adjust remedies in a feasible and practical way to eliminate the conditions or redress the injuries caused by unlawful action. Equitable remedies must be flexible if these underlying principles are to be enforced with fairness and precision. In this respect, as we observed in Swann, “a school desegregation case does not differ fundamentally from other cases involving the framing of equitable remedies to repair the denial of a constitutional right. The task is to correct, by a balancing of the individual and collective interests, the condition that offends the Constitution.” Swann, 402 U. S., at 15-16. The requirement of a unitary school system must be implemented according to this prescription.
Our application of these guiding principles in Pasadena Bd. of Education v. Spangler, 427 U. S. 424 (1976), is instructive. There we held that a District Court exceeded its remedial authority in requiring annual readjustment of school *488attendance zones in the Pasadena school district when changes in the racial makeup of the schools were caused by demographic shifts “not attributed to any segregative acts on the part of the [school district].” Id., at 436. In so holding we said:
“It may well be that petitioners have not yet totally achieved the unitary system contemplated by . - . . Swann. There has been, for example, dispute as to the petitioners’ compliance with those portions of the plan specifying procedures for hiring and promoting teachers and administrators. See 384 F. Supp. 846 (1974), vacated, 537 F. 2d 1031 (1976). But that does not undercut the force of the principle underlying the quoted language from Swann. In this case the District Court approved a plan designed to obtain racial neutrality in the attendance of students at Pasadena’s public schools. No one disputes that the initial implementation of this plan accomplished that objective. That being the case, the District Court was not entitled to require the [Pasadena Unified School District] to rearrange its attendance zones each year so as to ensure that the racial mix desired by the court was maintained in perpetuity. For having once implemented a racially neutral attendance pattern in order to remedy the perceived constitutional violations on the part of the defendants, the District Court had fully performed its function of providing the appropriate remedy for previous racially discriminatory attendance patterns.” Ibid.
See also id., at 438, n. 5 (“Counsel for the original plaintiffs has urged, in the courts below and before us, that the District Court’s perpetual ‘no majority of any minority’ requirement was valid and consistent with Swann, at least until the school system achieved ‘unitary’ status in all other respects such as the hiring and promoting of teachers and administrators. Since we have concluded that the case' is moot with *489regard to these plaintiffs, these arguments are not properly before us. It should be clear from what we have said that they have little substance”).
Today, we make explicit the rationale that was central in Spangler. A federal court in a school desegregation case has the discretion to order an incremental or partial withdrawal of its supervision and control. This discretion derives both from the constitutional authority which justified its intervention in the first instance and its ultimate objectives in formulating the decree. The authority of the court is invoked at the outset to remedy particular constitutional violations. In construing the remedial authority of the district courts, we have been guided by the principles that “judicial powers may be exercised only on the basis of a constitutional violation,” and that “the nature of the violation determines the scope of the remedy.” Swann, supra, at 16. A remedy is justifiable only insofar as it advances the ultimate objective of alleviating the initial constitutional violation.
We have said that the court’s end purpose must be to remedy the violation and, in addition, to restore state and local authorities to the control of a school system that is operating in compliance with the Constitution. Milliken v. Bradley, 433 U. S. 267, 280-281 (1977) (“[Tjhe federal courts in devising a remedy must take into account the interests of state and local authorities in managing their own affairs, consistent with the Constitution”). Partial relinquishment of judicial control, where justified by the facts of the case, can be an important and significant step in fulfilling the district court’s duty to return the operations and control of schools to local authorities. In Dowell, we emphasized that federal judicial supervision of local school systems was intended as a “temporary measure.” 498 U. S., at 247. Although this temporary measure has lasted decades, the ultimate objective has not changed — to return school districts to the control of local authorities. Just as a court has the obligation *490at the outset of a desegregation decree to structure a plan so that all available resources of the court are directed to comprehensive supervision of its decree, so too must a court provide an orderly means for withdrawing from control when it is shown that the school district has attained the requisite degree of compliance. A transition phase in which control is relinquished in a gradual way is an appropriate means to this end.
As we have long observed, “local autonomy of school districts is a vital national tradition.” Dayton Bd. of Education v. Brinkman, 433 U. S. 406, 410 (1977) (Dayton I). Returning schools to the control of local authorities at the earliest practicable date is essential to restore their true accountability in our governmental system. When the school district and all state entities participating with it in operating the schools make decisions in the absence of judicial supervision, they can be held accountable to the citizenry, to the political process, and to the courts in the ordinary course. As we discuss below, one of the prerequisites to relinquishment of control in whole or in part is that a school district has demonstrated its commitment to a course of action that gives full respect to the equal protection guarantees of the Constitution. Yet it must be acknowledged that the potential for discrimination and racial hostility is still present in our country, and its manifestations may emerge in new and subtle forms after the effects of de jure segregation have been eliminated. It is the duty of the State and its subdivisions to ensure that such forces do not shape or control the policies of its school systems. Where control lies, so too does responsibility.
We hold that, in the course of supervising desegregation plans, federal courts have the authority to relinquish supervision and control of school districts in incremental stages, before full compliance has been achieved in every area of school operations. While retaining jurisdiction over the case, the court may determine that it will not order further *491remedies in areas where the school district is in compliance with the decree. That is to say, upon a finding that a school system subject to a court-supervised desegregation plan is in compliance in some but not all areas, the court in appropriate cases may return control to the school system in those areas where compliance has been achieved, limiting further judicial supervision to operations that are not yet in full compliance with the court decree. In particular, the district court may determine that it will not order further remedies in the area of student assignments where racial imbalance is not traceable, in a proximate way, to constitutional violations.
A court’s discretion to order the incremental withdrawal of its supervision in a school desegregation case must be exercised in a manner consistent with the purposes and objectives of its equitable power. Among the factors which must inform the sound discretion of the court in ordering partial withdrawal are the following: whether there has been full and satisfactory compliance with the decree in those aspects of the system where supervision is to be withdrawn; whether retention of judicial control is necessary or practicable to achieve compliance with the decree in other facets of the school system; and whether the school district has demonstrated, to the public and to the parents and students of the once disfavored race, its good-faith commitment to the whole of the court’s decree and to those provisions of the law and the Constitution that were the predicate for judicial intervention in the first instance.
In considering these factors, a court should give particular attention to the school system’s record of compliance. A school system is better positioned to demonstrate its good-faith commitment to a constitutional course of action when its policies form a consistent pattern of lawful conduct directed to eliminating earlier violations. And, with the passage of time, the degree to which racial imbalances continue to represent vestiges of a constitutional violation may dimin*492ish, and the practicability and efficacy of various remedies can be evaluated with more precision.
These are the premises that guided our formulation in Dowell of the duties of a district court during the final phases of a desegregation case: “The District Court should address itself to whether the Board had complied in good faith with the desegregation decree since it was entered, and whether the vestiges of past discrimination had been eliminated to the extent practicable.” 498 U. S., at 249-250.
B
We reach now the question whether the Court of Appeals erred in prohibiting the District Court from returning to DCSS partial control over some of its affairs. We decide that the Court of Appeals did err in holding that, as a matter of law, the District Court had no discretion to permit DCSS to regain control over student assignment, transportation, physical facilities, and extracurricular activities, while retaining court supervision over the areas of faculty and administrative assignments and the quality of education, where full compliance had not been demonstrated.
It was an appropriate exercise of its discretion for the District Court to address the elements of a unitary system discussed in Green, to inquire whether other elements ought to be identified, and to determine whether minority students were being disadvantaged in ways that required the formulation of new and further remedies to ensure full compliance with the court’s decree. Both parties agreed that quality of education was a legitimate inquiry in determining DCSS’ compliance with the desegregation decree, and the trial court found it workable to consider the point in connection with its findings on resource allocation. Its order retaining supervision over this aspect of the case has not been challenged by the parties, and we need not examine it except as it underscores the school district’s record of compliance in some areas but not others. The District Court’s approach illustrates *493that the Green factors need not be a rigid framework. It illustrates also the uses of equitable discretion. By withdrawing control over areas where judicial supervision is no longer needed, a district court can concentrate both its own resources and those of the school district on the areas where the effects of de jure discrimination have not been eliminated and further action is necessary in order to provide real and tangible relief to minority students.
The Court of Appeals’ rejection of the District Court’s order rests on related premises: first, that given noncompliance in some discrete categories, there can be no partial withdrawal of judicial control; and second, until there is full compliance, heroic measures must be taken to ensure racial balance in student assignments system wide. Under our analysis and our precedents, neither premise is correct.
The Court of Appeals was mistaken in ruling that our opinion in Swann requires “awkward,” “inconvenient,” and “even bizarre” measures to achieve racial balance in student assignments in the late phases of carrying out a decree, when the imbalance is attributable neither to the prior de jure system nor to a later violation by the school district but rather to independent demographic-forces. In Swann we undertook to discuss the objectives of a comprehensive desegregation plan and the powers and techniques available to a district court in designing it at the outset. We confirmed that racial balance in school assignments was a necessary part of the remedy in the circumstances there presented. In the case before us the District Court designed a comprehensive plan for desegregation of DCSS in 1969, one that included racial balance in student assignments. The desegregation decree was designed to achieve maximum practicable desegregation. Its central remedy was the closing of black schools and the reassignment of pupils to neighborhood schools, with attendance zones that achieved racial balance. The plan accomplished its objective in the first year of operation, before dramatic demographic changes altered residen*494tial patterns. For the entire 17-year period respondents raised no substantial objection to the basic student assignment system, as the parties and the District Court concentrated on other mechanisms to eliminate the de jure taint.
That there was racial imbalance in student attendance zones was not tantamount to a showing that the school district was in noncompliance with the decree or with its duties under the law. Racial balance is not to be achieved for its own sake. It is to be pursued when racial imbalance has been caused by a constitutional violation. Once the racial imbalance due to the de jure violation has been remedied, the school district is under no duty to remedy imbalance that is caused by demographic factors. Swann, 402 U. S., at 31-32 (“Neither school authorities nor district courts are constitutionally required to make year-by-year adjustments of the racial composition of student bodies once the affirmative duty to desegregate has been accomplished and racial discrimination through official action is eliminated from the system. This does not mean that federal courts are without power to deal with future problems; but in the absence of a showing that either the school authorities or some other agency of the State has deliberately attempted to fix or alter demographic patterns to affect the racial composition of the schools, further intervention by a district court should not be necessary”). If the unlawful de jure policy of a school system has been the cause of the racial imbalance in student attendance, that condition must be remedied. The school district bears the burden of showing that any current imbalance is not traceable, in a proximate way, to the prior violation.
The findings of the District Court that the population changes which occurred in DeKalb County were not caused by the policies of the school district, but rather by independent factors, are consistent with the mobility that is a distinct characteristic of our society. In one year (from 1987 to 1988) over 40 million Americans, or 17.6% of the total population, *495moved households. U. S. Dept, of Commerce, Bureau of Census, Statistical Abstract of the United States 19 (111th ed. 1991) (Table 25). Over a third of those people moved to a different county, and over six million migrated between States. Ibid. In such a society it is inevitable that the demographic makeup of school districts, based as they are on political subdivisions such as counties and municipalities, may undergo rapid change.
The effect of changing residential patterns on the racial composition of schools, though not always fortunate, is somewhat predictable. Studies show a high correlation between residential segregation and school segregation. Wilson & Taeuber, Residential and School Segregation: Some Tests of Their Association, in Demography and Ethnic Groups 57-58 (P. Bean & W. Frisbie eds. 1978). The District Court in this case heard evidence tending to show that racially stable neighborhoods are not likely to emerge because whites prefer a racial mix of 80% white and 20% black, while blacks prefer a 50-50 mix.
Where resegregation is a product not of state action but of private choices, it does not have constitutional implications. It is beyond the authority and beyond the practical ability of the federal courts to try to counteract these kinds of continuous and massive demographic shifts. To attempt such results would require ongoing and never-ending supervision by the courts of school districts simply because they were once de jure segregated. Residential housing choices, and their attendant effects on the racial composition of schools, present an ever-changing pattern, one difficult to address through judicial remedies.
In one sense of the term, vestiges of past segregation by state decree do remain in our society and in our schools. Past wrongs to the black race, wrongs committed by the State and in its name, are a stubborn fact of history. And stubborn facts of history linger and persist. But though we cannot escape our history, neither must we overstate its con*496sequences in fixing legal responsibilities. The vestiges of segregation that are the concern of the law in a school case may be subtle and intangible but nonetheless they must be so real that they have a causal link to the de jure violation being remedied. It is simply not always the case that demographic forces causing population change bear any real and substantial relation to a de jure violation. And the law need not proceed on that premise.
As the de jure violation becomes moré remote in time and these demographic changes intervene, it becomes less likely that a current racial imbalance in a school district is a vestige of the prior de jure system. The causal link between current conditions and the prior violation is even more attenuated if the school district has demonstrated its good faith. In light of its finding that the demographic changes in DeKalb County are unrelated to the prior violation, the District Court was correct to entertain the suggestion that DCSS had no duty to achieve system-wide racial balance in the student population. It was appropriate for the District Court to examine the reasons for the racial imbalance before ordering an impractical, and no doubt massive, expenditure of funds to achieve racial balance after 17 years of efforts to implement the comprehensive plan in a district where there were fundamental changes in demographics, changes not attributable to the former de jure regime or any later actions by school officials. The District Court’s determination to order instead the expenditure of scarce resources in areas such as the quality of education, where full compliance had not yet been achieved, underscores the uses of discretion in framing equitable remedies.
To say, as did the Court of Appeals, that a school district must meet all six Green factors before the trial court can declare the system unitary and relinquish its control over school attendance zones, and to hold further that racial balancing by all necessary means is required in the interim, is *497simply to vindicate a legal phrase. The law is not so formalistic. A proper rule must be based on the necessity to find a feasible remedy that ensures system-wide compliance with the court decree and that is directed to curing the effects of the specific violation.
We next consider whether retention of judicial control over student attendance is necessary or practicable to achieve compliance in other facets of the school system. Racial balancing in elementary and secondary school student assignments may be a legitimate remedial device to correct other fundamental inequities that were themselves caused by the constitutional violation. We have long recognized that the Green factors may be related or interdependent. Two or more Green factors may be intertwined or synergistic in their relation, so that a constitutional violation in one area cannot be eliminated unless the judicial remedy addresses other matters as well. We have observed, for example, that student segregation and faculty segregation are often related problems. See Dayton Bd. of Education v. Brinkman, 443 U.S. 526, 536 (1979) (Dayton II) (“‘[Purposeful segregation of faculty by race was inextricably tied to racially motivated student assignment practices”’); Rogers v. Paul, 382 U. S. 198, 200 (1965) (students have standing to challenge racial allocation of faculty because “racial allocation of faculty denies them equality of educational opportunity without regard to segregation of pupils”). As a consequence, a continuing violation in one area may need to be addressed by remedies in another. See, e. g., Bradley v. Richmond School Bd., 382 U. S. 103, 105 (1965) (per curiam) (“There is no merit to the suggestion that the relation between faculty allocation on an alleged racial basis and the adequacy of the desegregation plans is entirely speculative”); Vaughns v. Board of Education of Prince George’s County, 742 F. Supp. 1275, 1291 (Md. 1990) (“[T]he components of *498a school desegregation plan are interdependent upon, and interact with, one another, so that changes with respect to one component may impinge upon the success or failure of another”).
There was no showing that racial balancing was an appropriate mechanism to cure other deficiencies in this case. It is true that the school district was not in compliance with respect to faculty assignments, but the record does not show that student reassignments would be a feasible or practicable way to remedy this defect. To the contrary, the District Court suggests that DCSS could solve the faculty assignment problem by reassigning a few teachers per school. The District Court, not having our analysis before it, did not have the opportunity to make specific findings and conclusions on this aspect of the case, however. Further proceedings are appropriate for this purpose.
The requirement that the school district show its good-faith commitment to the entirety of a desegregation plan so that parents, students, and the public have assurance against further injuries or stigma also should be a subject for more specific findings. We stated in Dowell that the good-faith compliance of the district with the court order over a reasonable period of time is a factor to be considered in deciding whether or not jurisdiction could be relinquished. 498 U. S., at 249-250 (“The District Court should address itself to whether the Board had complied in good faith with the desegregation decree since it was entered, and whether the vestiges of past discrimination had been eliminated to the extent practicable”). A history of good-faith compliance is evidence that any current racial imbalance is not the product of a new de jure violation, and enables the district court to accept the school board’s representation that it has accepted the principle of racial equality and will not suffer intentional discrimination in the future. See Morgan v. Nucci, 831 *499F. 2d, at 321 (“A finding of good faith... reduces the possibility that a school system’s compliance with court orders is but a temporary constitutional ritual”).
When a school district has not demonstrated good faith under a comprehensive plan to remedy ongoing violations, we have without hesitation approved comprehensive and continued district court supervision. See Columbus Bd. of Education v. Penick, 443 U. S. 449, 461 (1979) (predicating liability in part on the finding that the school board “ ‘never actively set out to dismantle [the] dual system,’ ” Penick v. Columbus Bd. of Education, 429 F. Supp. 229, 260 (SD Ohio 1977)); Dayton II, supra, at 534 (adopting Court of Appeals holding that the “intentionally segregative impact of various practices since 1954 . . . were of systemwide import and an appropriate basis for a systemwide remedy”).
In contrast to the circumstances in Penick and Brinkman, the District Court in this case stated that throughout the period of judicial supervision it has been impressed by the successes DCSS has achieved and its dedication to providing a quality education for all students, and that DCSS “has travelled the often long road to unitary status almost to its end.” With respect to those areas where compliance had not been achieved, the District Court did not find that DCSS had acted in bad faith or engaged in further acts of discrimination since the desegregation plan went into effect. This, though, may not be the equivalent of a finding that the school district has an affirmative commitment to comply in good faith with the entirety of a desegregation plan, and further proceedings are appropriate for this purpose as well.
The judgment is reversed, and the case is remanded to the Court of Appeals. It should determine what issues are open for its further consideration in light of the previous briefs and arguments of the parties and in light of the principles set forth in this opinion. Thereupon it should order further *500proceedings as necessary or order an appropriate remand to the District Court.
Each party is to bear its own costs.
It is so ordered.
Justice Thomas took no part in the consideration or decision of this case.