delivered the opinion of the Court.
As this school desegregation litigation enters its 18th year, we are called upon again to review the decisions of the lower courts. In this case, the State of Missouri has challenged the District Court’s order of salary increases for virtually all instructional and noninstructional staff within the Kansas City, Missouri, School District (KCMSD) and the District Court’s order requiring the State to continue to fund remedial “quality education” programs because student achievement levels were still “at or below national norms at many grade levels.”
*74I
A general overview of this litigation is necessary for proper resolution of the issues upon which we granted cer-tiorari. This case has been before the same United States District Judge since 1977. Missouri v. Jenkins, 491 U. S. 274, 276 (1989) (Jenkins I). In that year, the KCMSD, the school board, and the children of two school board members brought suit against the State and other defendants. Plaintiffs alleged that the State, the surrounding suburban school districts (SSD’s), and various federal agencies had caused and perpetuated a system of racial segregation in the schools of the Kansas City metropolitan area. The District Court realigned the KCMSD as a nominal defendant and certified as a class, present and future KCMSD students. The KCMSD brought a cross-claim against the State for its failure to eliminate the vestiges of its prior dual school system.
After a trial that lasted Tk months, the District Court dismissed the case against the federal defendants and the SSD’s, but determined that the State and the KCMSD were liable for an intradistrict violation, i e., they had operated a segregated school system within the KCMSD. Jenkins v. Missouri, 593 F. Supp. 1485 (WD Mo. 1984). The District Court determined that prior to 1954 “Missouri mandated segregated schools for black and white children.” Id., at 1490. Furthermore, the KCMSD and the State had failed in their affirmative obligations to eliminate the vestiges of the State’s dual school system within the KCMSD. Id., at 1504.
In June 1985, the District Court issued its first remedial order and established as its goal the “elimination of all vestiges of state imposed segregation.” Jenkins v. Missouri, 639 F. Supp. 19, 23 (WD Mo. 1985). The District Court determined that “[segregation ha[d] caused a system wide reduction in student achievement in the schools of the KCMSD.” Id., at 24. The District Court made no particularized findings regarding the extent that student achieve*75ment had been reduced or what portion of that reduction was attributable to segregation. The District Court also identified 25 schools within the KCMSD that had enrollments of 90% or more black students. Id., at 36.
The District Court, pursuant to plans submitted by the KCMSD and the State, ordered a wide range of quality education programs for all students attending the KCMSD. First, the District Court ordered that the KCMSD be restored to an AAA classification, the highest classification awarded by the State Board of Education. Id., at 26. Second, it ordered that the number of students per class be reduced so that the student-to-teacher ratio was below the level required for AAA standing. Id., at 28-29. The District Court justified its reduction in class size as
“an essential part of any plan to remedy the vestiges of segregation in the KCMSD. Reducing class size will serve to remedy the vestiges of past segregation by increasing individual attention and instruction, as well as increasing the potential for desegregative educational experiences for KCMSD students by maintaining and attracting non-minority enrollment.” Id., at 29.
The District Court also ordered programs to expand educational opportunities for all KCMSD students: full-day kindergarten; expanded summer school; before- and after-school tutoring; and an early childhood development program. Id., at 30-33. Finally, the District Court implemented a state-funded “effective schools” program that consisted of substantial yearly cash grants to each of the schools within the KCMSD. Id., at 33-34. Under the “effective schools” program, the State was required to fund programs at both the 25 racially identifiable schools as well as the 43 other schools within the KCMSD. Id., at 33.
The KCMSD was awarded an AAA rating in the 1987-1988 school year, and there is no dispute that since that time it has “‘maintained and greatly exceeded AAA require*76ments.’” 19 F. 3d 393, 401 (CA8 1994) (Beam, J., dissenting from denial of rehearing en banc). The total cost for these quality education programs has exceeded $220 million. Missouri Department of Elementary and Secondary Education, KCMSD Total Desegregation Program Expenditures (Sept. 30, 1994) (Desegregation Expenditures).
The District Court also set out to desegregate the KCMSD but believed that “[t]o accomplish desegregation within the boundary lines of a school district whose enrollment remains 68.3% black is a difficult .task.” 639 F. Supp., at 38. Because it had found no interdistrict violation, the District Court could not order mandatory interdistrict redistribution of students between the KCMSD and the surrounding SSD’s. Ibid.; see also Milliken v. Bradley, 418 U. S. 717 (1974) (Milliken I). The District Court refused to order additional mandatory student reassignments because they would “increase the instability of the KCMSD and reduce the potential for desegregation.” 639 F. Supp., at 38. Relying on favorable precedent from the Eighth Circuit, the District Court determined that “[ajchievement of AAA status, improvement of the quality of education being offered at the KCMSD schools, magnet schools, as well as other components of this desegregation plan can serve to maintain and hopefully attract non-minority student enrollment.” Ibid.
In November 1986, the District Court approved a compre-hensivé magnet school and capital improvements plan and held the State and the KCMSD jointly and severally liable for its funding. 1 App. 130-193. Under the District Court’s plan, every senior high school, every middle school, and one-half of the elementary schools were converted into magnet schools.1 Id., at 131. The District Court adopted the *77magnet-school program to “provide a greater educational opportunity to all KCMSD students,” id., at 131-132, and because it believed “that the proposed magnet plan [was] so attractive that it would draw non-minority students from the private schools who have abandoned or avoided the KCMSD, and draw in additional non-minority students from the suburbs.” Id., at 132. The District Court felt that “[t]he long-term benefit of all KCMSD students of a greater educational opportunity in an integrated environment is worthy of such an investment.” Id., at 133. Since its inception, the magnet-school program has operated at a cost, including magnet transportation, in excess of $448 million. See Desegregation Expenditures. In April 1993, the District Court considered, but ultimately rejected, the plaintiffs’ and the KCMSD’s proposal seeking approval of a long-range magnet renewal program that included a 10-year budget of well over $500 million, funded by the State and the KCMSD on a joint-and-several basis. App. to Pet. for Cert. A-123.
In June 1985, the District Court ordered substantial capital improvements to combat the deterioration of the KCMSD’s facilities. In formulating its capital-improvements plan, the District Court dismissed as “irrelevant” the “State’s argument that the present condition of the facilities [was] not traceable to unlawful segregation.” 639 F. Supp., at 40. Instead, the District Court focused on- its responsibility to “remed[y] the vestiges of segregation” and to “implement] a desegregation plan which w[ould] maintain and attract non-minority enrollment.” Id., at 41. The initial phase of the capital-improvements plan cost $37 million. Ibid. The District Court also required the KCMSD to present further capital-improvements proposals “in order to bring its facilities to a point comparable with the facilities in neighboring suburban school districts.” Ibid. In November 1986, the District Court approved further capital improvements in order to remove the vestiges of racial segre*78gation and “to ... attract non-minority students back to the KCMSD.” App. to Pet. for Cert. A-133 to A-134.
In September 1987, the District Court adopted, for the most part, KCMSD’s long-range capital-improvements plan at a cost in excess of $187 million. Jenkins v. Missouri, 672 F. Supp. 400, 408 (WD Mo. 1987). The plan called for the renovation of approximately 55 schools, the closure of 18 facilities, and the construction of 17 new schools. Id., at 405. The District Court rejected what it referred to as the “ ‘patch and repair’ approach proposed by the State” because it “would not achieve suburban comparability or the visual attractiveness sought by the Court as it would result in floor coverings with unsightly sections of mismatched carpeting and tile, and individual walls possessing different shades of paint.” Id., at 404. The District Court reasoned that “if the KCMSD schools underwent the limited renovation proposed by the State, the schools would continue to be unattractive and substandard, and would certainly serve as a deterrent to parents considering enrolling their children in KCMSD schools.” Id., at 405. As of 1990, the District Court had ordered $260 million in capital improvements. Missouri v. Jenkins, 495 U. S. 33, 61 (1990) (Jenkins II) (Kennedy, J., concurring in part and concurring in judgment). Since then, the total cost of capital improvements ordered has soared to over $540 million.
As part of its desegregation plan, the District Court has ordered salary assistance to the KCMSD. In 1987, the District Court initially ordered salary assistance only for teachers within the KCMSD. Since that time, however, the District Court has ordered salary assistance to all but three of the approximately 5,000 KCMSD employees. The total cost of this component of the desegregation remedy since 1987 is over $200 million. See Desegregation Expenditures.
The District Court’s desegregation plan has been described as the most ambitious and expensive remedial program in the history of school desegregation. 19 F. 3d, at 397 *79(Beam, J., dissenting from denial of rehearing en banc). The annual cost per pupil at the KCMSD far exceeds that of the neighboring SSD’s or of any school district in Missouri. Nevertheless, the KCMSD, which has pursued a “friendly adversary” relationship with the plaintiffs, has continued to propose ever more expensive programs.2 As a result, the desegregation costs have escalated and now are approaching an annual cost of $200 million. These massive expenditures have financed
“high schools in which every classroom will have air conditioning, an alarm system, and 15 microcomputers; a 2,000-square-foot planetarium; green houses and vivariums; a 25-acre farm with an air-conditioned meeting room for 104 people; a Model United Nations wired for language translation; broadcast capable radio and television studios with an editing and animation lab; a temperature controlled art gallery; movie editing and screening rooms; a 3,500-square-foot dust-free diesel mechanics room; 1,875-square-foot elementary school animal rooms for use in a zoo project; swimming pools; and numerous other facilities.” Jenkins II, 495 U. S., at 77 (Kennedy, J., concurring in part and concurring in judgment).
Not surprisingly, the cost of this remedial plan has “far exceeded KCMSD’s budget, or for that matter, its authority to tax.” Id., at 60. The State, through the operation of joint- and-several liability, has borne the brunt of these costs. The District Court candidly has acknowledged that it has “allowed the District planners to dream” and “provided the *80mechanism for th[ose] dreams to be realized.” App. to Pet. for Cert. A-133. In short, the District Court “has gone to great lengths to provide KCMSD with facilities and opportunities not available anywhere else in the country.” Id., at
II
With this background, we turn to the present controversy. First, the State has challenged the District Court’s requirement that it fund salary increases for KCMSD instructional and noninstructional staff. Id., at A-76 to A-93 (District Court’s Order of June 15, 1992); id., at A-94 to A-109 (District Court’s Order of June 30, 1993); id., at A-110 to A-121 (District Court’s Order of July 30,1993). The State claimed that funding for salaries was beyond the scope of the District Court’s remedial authority. Id., at A-86. Second, the State has challenged the District Court’s order requiring it to continue to fund the remedial quality education programs for the 1992-1993 school year. Id., at A-69 to A-75 (District Court’s Order of June 17, 1992). The State contended that under Freeman v. Pitts, 503 U. S. 467 (1992), it had achieved partial unitary status with respect to the quality education programs already in place. As a result, the State argued that the District Court should have relieved it of responsibility for funding those programs.
The District Court rejected the State’s arguments. It first determined that the salary increases were warranted because “[h]igh quality personnel are necessary not only to implement specialized desegregation programs intended to ‘improve educational opportunities and reduce racial isolation’, but also to ‘ensure that there is no diminution in the quality of its regular academic program.’ ” App. to Pet. for Cert. A-87 (citations omitted). Its “ruling [was] grounded in remedying the vestiges of segregation by improving the desegregative attractiveness of the KCMSD.” Id., at A-90. The District Court did not address the State’s Freeman arguments; nevertheless, it ordered the State to continue to *81fund the quality education programs for the 1992-1993 school year. See App. to Pet. for Cert. A-70.
The Court of Appeals for the Eighth Circuit affirmed. 11 F. 3d 755 (1993). It rejected the State’s argument that the salary increases did not directly address and relate to the State’s constitutional violation and that “low teacher salaries d[id] not flow from any earlier constitutional violations by the State.” Id., at 767. In doing so, it observed that “[i]n addition to compensating the victims, the remedy in this case was also designed to reverse white flight by offering superior educational opportunities.” Ibid.; see also 13 F. 3d 1170, 1172 (1993) (affirming the District Court’s June 30, 1993, and July 30, 1993, orders).
The Court of Appeals concluded that the District Court implicitly had rejected the State’s Freeman arguments in spite of the fact that it had failed “to articulate . . . even a eonclusory rejection” of them. 11 F. 3d, at 765. It looked to the District Court’s comments from the bench and its later orders to “illuminate the June 1992 order.” Id., at 761. The Court of Appeals relied on statements made by the District Court during a May 28, 1992, hearing:
“The Court’s goal was to integrate the Kansas City, Missouri, School District to the maximum degree possible, and all these other matters were elements to be used to try to integrate the Kansas City, Missouri, schools so the goal is integration. That’s the goal. And a high standard of quality education. The magnet schools, the summer school program and all these programs are tied to that goal, and until such time as that goal has been reached, then we have not reached the goal.... The goal is to integrate the Kansas City, Missouri, School district. So I think we are wasting our time.” 2 App. 482 (emphasis added).
See 11 F. 3d, at 761. Apparently, the Court of Appeals extrapolated from the findings regarding the magnet-school *82program and later orders and imported those findings wholesale to reject the State’s request for a determination of partial unitary status as to the quality education programs. See id., at 761-762. It found significant the District Court’s determination that although “there had been a trend of improvement in academic achievement, . . . the school district was far from reaching its maximum potential because KCMSD is still at or below national norms at many grade levels.” Ibid. It went on to say that with respect to quality education, “implementation of programs in and of itself is not sufficient. The test, after all, is whether the vestiges of segregation, here the systemwide reduction in student achievement, have been eliminated to the greatest extent practicable. The success of quality of education programs must be measured by their effect on the students, particularly those who have been the victims of segregation.” Id., at 766.
The Court of Appeals denied rehearing en banc, with five judges dissenting. 19 F. 3d, at 395. The dissent first examined the salary increases ordered by the District Court and characterized “the current effort by the KCMSD and the American Federation of Teachers .. . aided by the plaintiffs, to bypass the collective bargaining process” as “uncalled for” and “probably not an exercise reasonably related to the constitutional violations found by the court.” Id., at 399. The dissent also “agree[d] with the [S]tate that logic d[id] not directly relate the pay of parking lot attendants, trash haulers and food handlers ... to any facet or phase of the desegregation plan or to the constitutional violations.” Ibid.
Second, the dissent believed that in evaluating whether the KCMSD had achieved partial unitary status in its quality education programs, the District Court and the panel had
“misrea[d] Freeman and create[d] a hurdle to the withdrawal of judicial intervention from public education that has no support in the law. The district court has, *83with the approbation of the panel, imbedded a student achievement goal measured by annual standardized tests into its test of whether the KCMSD has built a high-quality educational system sufficient to remedy past discrimination. The Constitution requires no such standard.” Id., at 400.
The dissent noted that “KCMSD students have in place a system that offers more educational opportunity than anywhere in America,” id., at 403, but that the District Court was “‘not satisfied that the District has reached anywhere close to its maximum potential because the District is still at or below national norms át many grade levels,’ ” ibid, (emphasis added). The dissent concluded that this case, “as it now proceeds, involves an exercise in pedagogical sociology, not constitutional adjudication.” Id., at 404.
Because of the importance of the issues, we granted certio-rari to consider the following: (1) whether the District Court exceeded its constitutional authority when it granted salary increases to virtually all instructional and noninstructional employees of the KCMSD, and (2) whether the District Court properly relied upon the fact that student achievement test scores had failed to rise to some unspecified level when it declined to find that the State had achieved partial unitary status as to the quality education programs. 512 U. S. 1287 (1994).
Ill
Respondents argue that the State may no longer challenge the District Court’s remedy, and in any event, the propriety of the remedy is not before the Court. Brief for Respondents KCMSD et al. 40-49; Brief for Respondents Jenkins et al. 23. We disagree on both counts. In Jenkins II, we granted certiorari to review the manner in which the District Court had funded this desegregation remedy. 495 U. S., at 37. Because we had denied certiorari on the State’s *84challenge to review the scope of the remedial order, we resisted the State’s efforts to challenge the scope of the remedy. Id., at 53; cf. id., at 80 (Kennedy, J., concurring in part and concurring in judgment). Thus, we neither “approved]” nor “disapproved] the Court of Appeals’ conclusion that the District Court’s remedy was proper.” Id., at 53.
Here, however, the State has challenged the District Court’s approval of across-the-board salary increases for instructional and noninstructional employees as an action beyond its remedial authority. Pet. for Cert, i.3 An analysis of the permissible scope of the District Court’s remedial authority is necessary for a proper determination of whether the order of salary increases is beyond the District Court’s remedial authority, see Milliken I, 418 U. S., at 738-740, 745, and thus, it is an issue subsidiary to our ultimate inquiry. Cf. Yee v. Escondido, 503 U. S. 519, 537 (1992). Given that the District Court’s basis for its salary order was grounded in “improving the desegregative attractiveness of the KCMSD,” App. to Pet. for Cert. A-90, we must consider the propriety of that reliance in order to resolve properly the State’s challenge to that order. We conclude that a challenge to the scope of the District Court’s remedy is fairly included in the question presented. See this Court’s Rule 14.1; Procunier v. Navarette, 434 U. S. 555, 560, n. 6 (1978) (“Since consideration of these issues is essential to analysis of the Court of Appeals’ [decision] we shall also treat these questions as subsidiary issues ‘fairly comprised’ by the question presented”); see also United States v. Mendenhall, 446 U. S. 544, 551-552, n. 5 (1980) (opinion of Stewart, J.) (Where *85the determination of a question “is essential to the correct disposition of the other issues in the case, we shall treat it as ‘fairly comprised’ by the questions presented in the petition for certiorari”); cf. Yee, supra, at 536-537.
Justice Souter argues that our decision to review the scope of the District Court’s remedial authority is both unfair and imprudent. Post, at 147. He claims that factors such as our failure to grant certiorari on the State’s challenge to the District Court’s remedial authority in 1988 “lulled [respondents] into addressing the case without sufficient attention to the foundational issue, and their lack of attention has now infected the Court’s decision.” Post, at 139. Justice Souter concludes that we have “decide[d] the issue without any warning to respondents.” Post, at 147. These arguments are incorrect.
Of course, “[t]he denial of a writ of certiorari imports no expression of opinion upon the merits of the case, as the bar has been told many times.” United States v. Carver, 260 U. S. 482, 490 (1923). A fortiori, far from lulling respondents into a false sense of security, our previous decision in Jenkins v. Missouri put respondents on notice that the Court had not affirmed the validity of the District Court’s remedy, 495 U. S., at 53, and that at least four Justices of the Court questioned that remedy, id., at 75-80 (Kennedy, J., concurring in part and concurring in judgment).
With respect to the specific orders at issue here, the State has once again challenged the scope of the District Court’s remedial authority. The District Court was aware of this fact. See App. to Pet. for Cert. A-86 (“The State claims that the Court should not approve desegregation funding for salaries because such funding would be beyond the scope of the Court’s remedial authority”) (District Court’s June 25, 1992, order); id., at A-97 (“The State has argued repeatedly and currently on appeal that the salary component is not a valid component of the desegregation remedy”) (District *86Court’s June 30, 1993, order). The Court of Appeals also understood that the State had renewed this challenge. See 11 F. 3d, at 766 (“The State argues first that the salary increase remedy sought exceeded that necessary to remedy the constitutional violations, and alternatively, that if the district court had lawful authority to impose the increases, it abused its discretion in doing so”); id., at 767 (“The State’s legal argument is that the district court should have denied the salary increase funding because it is contrary to Milliken [v. Bradley, 433 U. S. 267 (1977),] and Swann [v. Charlotte-Mecklenburg Bd. of Ed., 402 U. S. 1 (1971),] in that it does not directly address and relate to the State’s constitutional violation”); 13 F. 3d, at 1172 (“We reject the State’s argument that the salary order is contrary to Milliken II and Swann”). The State renewed this same challenge in its petition for certiorari, Pet. for Cert, i, and argued here that the District Court’s salary orders were beyond the scope of its remedial authority. Brief for Petitioners 27-32; Reply Brief for Petitioners 6-12. In the 100 pages of briefing provided by respondents, they have argued that the State’s challenge to the scope of the District Court’s remedial authority is not fairly presented and is meritless. See Brief for Respondents KCMSD et al. 40-49; Brief for Respondents Jenkins et al. 2-21, 44-49; cf. Reply Brief for Petitioners 2 (“[Respondents . . . urge the Court to dismiss the writ as improvidently granted. This is not surprising; respondents cannot defend the excesses of the courts below”).
In short, the State has challenged the scope of the District Court’s remedial authority. The District Court, the Court of Appeals, and respondents have recognized this to be the case. Contrary to Justice Souter’s arguments, there is no unfairness or imprudence in deciding issues that have been passed upon below, are properly before us, and have been briefed by the parties. We turn to the questions presented.
Almost 25 years ago, in Swann v. Charlotte-Mecklenburg Bd. of Ed., 402 U. S. 1 (1971), we dealt with the authority of a district court to fashion remedies for a school district that *87had been segregated in law in violation of the Equal Protection Clause of the Fourteenth Amendment. Although recognizing the discretion that must necessarily adhere in a district court in fashioning a remedy, we also recognized the limits on such remedial power:
“[Elimination of racial discrimination in public schools is a large task and one that should not be retarded by efforts to achieve broader purposes lying beyond the jurisdiction of the school authorities. One vehicle can carry only a limited amount of baggage. It would not serve the important objective of Brown [v. Board of Education, 347 U. S. 483 (1954),] to seek to use school desegregation cases for purposes beyond their scope, although desegregation of schools ultimately will have impact on other forms of discrimination.” Id., at 22-23.
Three years later, in Milliken I, 418 U. S. 717 (1974), we held that a District Court had exceeded its authority in fashioning interdistrict relief where the surrounding school districts had not themselves been guilty of any constitutional violation. Id., at 746-747. We said that a desegregation remedy “is necessarily designed, as all remedies are, to restore the victims of discriminatory conduct to the position they would have occupied in the absence of such conduct.” Id., at 746. “[WJithout an interdistrict violation and inter-district effect, there is no constitutional wrong calling for an interdistrict remedy.” Id., at 745. We also rejected “[t]he suggestion ... that schools which have a majority of Negro students are not ‘desegregated,’ whatever the makeup of the school district’s population and however neutrally the district lines have been drawn and administered.” Id., at 747, n. 22; see also Freeman, 503 U. S., at 474 (“[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”).
*88Three years later, in Milliken v. Bradley, 433 U. S. 267 (1977) (Milliken II), we articulated a three-part framework derived from our prior cases to guide district courts in the exercise of their remedial authority.
“Ini the first place, like other equitable remedies, the nature of the desegregation remedy is to be determined by the nature and scope of the constitutional violation. Swann v. Charlotte-Mecklenburg Board of Education, 402 U. S., at 16. The remedy must therefore be related to ‘the condition alleged to offend the Constitution.. ..’ Milliken I, 418 U. S., at 738. Second, the decree must indeed be remedial in nature, that is, it must be designed as nearly as possible ‘to restore the victims of discriminatory conduct to the position they would have occupied in the absence of such conduct.’ Id., at 746. Third, the 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.” Id., at 280-281 (footnotes omitted).
We added that the “principle that the nature and scope of the remedy are to be determined by the violation means simply that federal-court decrees must directly address and relate to the constitutional violation itself.” Id., at 281-282. In applying these principles, we have identified “student assignments, . . . ‘faculty, staff, transportation, extracurricular activities and facilities’ ” as the most important indicia of a racially segregated school system. Board of Ed. of Oklahoma City Public Schools v. Dowell, 498 U. S. 237, 250 (1991) (quoting Green v. School Bd. of New Kent Cty., 391 U. S. 430, 435 (1968)).
Because “federal supervision of local school systems was intended as a temporary measure to remedy past discrimination,” Dowell, supra, at 247, we also have considered the showing that must be made by a school district operating under a desegregation order for complete or partial relief from that order. In Freeman, we stated that
*89“[a]mong the factors which must inform the sound discretion of the court in ordering partial withdrawal are the following: [1] whether there has been full and satisfactory compliance with the decree in those aspects of the system where supervision is to be withdrawn; [2] whether retention of judicial control is necessary or practicable to achieve compliance with the decree in other facets of the school system; and [3] 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 courts’ decree and to those provisions of the law and the Constitution that were the predicate for judicial intervention in the first instance.” 503 U. S., at 491.
The ultimate inquiry is “‘whether the [constitutional violator] ha[s] complied in good faith with the desegregation decree since it was entered, and whether the vestiges of past discrimination ha[ve] been eliminated to the extent practicable.’” Id., at 492 (quoting Dowell, supra, at 249-250).
Proper analysis of the District Court’s orders challenged here, then, must rest upon their serving as proper means to the end of restoring the victims of discriminatory conduct to the position they would have occupied in the absence of that conduct and their eventual restoration of “state and local authorities to the control of a school system that is operating in compliance with the Constitution.” 503 U. S., at 489. We turn to that analysis.
The State argues that the order approving salary increases is beyond the District Court’s authority because it was crafted to serve an “interdistrict goal,” in spite of the fact that the constitutional violation in this case is “intradis-trict” in nature. Brief for Petitioners 19. “[T]he nature of the desegregation remedy is to be determined by the nature and scope of the constitutional violation.” Milliken II, supra, at 280; Pasadena City Bd. of Ed. v. Spangler, 427 U. S. 424, 434 (1976) (“ ‘[T]here are limits’ beyond which a *90court may not go in seeking to dismantle a dual school system”). The proper response to an intradistrict violation is an intradistrict remedy, see Milliken I, supra, at 746-747; Milliken II, supra, at 280, that serves to eliminate the racial identity of the schools within the affected school district by eliminating, as far as practicable, the vestiges of de jure segregation in all facets of their operations. See Dowell, supra, at 250; see also Swann, 402 U. S., at 18-19; Green, supra, at 435.
Here, the District Court has found, and the Court of Appeals has affirmed, that this case involved no interdistrict constitutional violation that would support interdistrict relief. Jenkins II, 495 U. S., at 37, n. 3 (“The District Court also found that none of the alleged discriminatory actions had resulted in lingering interdistrict effects and so dismissed the suburban school districts and denied interdistrict relief”); id., at 76 (Kennedy, J., concurring in part and concurring in judgment) (“[T]here was no interdistrict constitutional violation that would support mandatory interdistrict relief”).4 Thus, the proper response by the District Court should have been to eliminate to the extent practicable the vestiges of prior de jure segregation within the KCMSD: a systemwide reduction in student achievement and the existence of 25 racially identifiable schools with a population of over 90% black students. 639 F. Supp., at 24, 36.
*91The District Court and Court of Appeals, however, have felt that because the KCMSD’s enrollment remained 68.3% black, a purely miradistrict remedy would be insufficient. Id., at 38; Jenkins v. Missouri, 865 F. 2d 1296, 1302 (CA8 1988) (“[V]oluntary interdistrict remedies may be used to make meaningful integration possible in a predominantly minority district”). But, as noted in Milliken I, 418 U. S. 717 (1974), we have rejected the suggestion “that schools which have a majority of Negro students are not ‘desegregated’ whatever the racial makeup of the school district’s population and however neutrally the district lines have been drawn and administered.” Id., at 747, n. 22; see Milliken II, 433 U. S., at 280, n. 14 (“[T]he Court has consistently held that the Constitution is not violated by racial imbalance in the schools, without more”); Spangler, supra, at 434.5
Instead of seeking to remove the racial identity of the various schools within the KCMSD, the District Court has set out on a program to create a school district that was equal to or superior to the surrounding SSD’s. Its remedy has focused on “desegregative attractiveness,” coupled with “suburban comparability.” Examination of the District Court’s reliance on “desegregative attractiveness” and “suburban comparability” is instructive for our ultimate resolution of the salary-order issue.
The purpose of desegregative attractiveness has been not only to remedy the systemwide reduction in student achievement, but also to attract nonminority students not presently enrolled in the KCMSD. This remedy has included an elaborate program of capital improvements, course enrichment, *92and extracurricular enhancement not simply in the formerly identifiable black schools, but in schools throughout the district. The District Court’s remedial orders have converted every senior high school, every middle school, and one-half of the elementary schools in the KCMSD into “magnet” schools. The District Court’s remedial order has all but made the KCMSD itself into a magnet district.
We previously have approved of intradistrict desegregation remedies involving magnet schools. See, e. g., Milliken II, supra, at 272. Magnet schools have the advantage of encouraging voluntary movement of students within a school district in a pattern that aids desegregation on a voluntary basis, without requiring extensive busing and redrawing of district boundary lines. Cf. Jenkins II, supra, at 59-60 (Kennedy, J., concurring in part and concurring in judgment) (citing Milliken II, supra, at 272). As a component in an intradistrict remedy, magnet schools also are attractive because they promote desegregation while limiting the withdrawal of white student enrollment that may result from mandatory student reassignment. See 639 F. Supp., at 37; cf. United States v. Scotland Neck City Bd. of Ed., 407 U. S. 484, 491 (1972).
The District Court’s remedial plan in this case, however, is not designed solely to redistribute the students within the KCMSD in order to eliminate racially identifiable schools within the KCMSD. Instead, its purpose is to attract non-minority students from outside the KCMSD schools. But this mierdistrict goal is beyond the scope of the miradistrict violation identified by the District Court. In effect, the District Court has devised a remedy to accomplish indirectly what it admittedly lacks the remedial authority to mandate directly: the interdistrict transfer of students. 639 F. Supp., at 38 (‘“[BJecause of restrictions on this Court’s remedial powers in restructuring the operations of local and state government entities,’ any mandatory plan which would go beyond the boundary lines of KCMSD goes far beyond *93the nature and extent of the constitutional violation [that] this Court found existed”).
In Milliken I we determined that a desegregation remedy that would require mandatory interdistrict reassignment of students throughout the Detroit metropolitan area was an impermissible interdistriet response to the intradistrict violation identified. 418 U. S., at 745. In that case, the lower courts had ordered an interdistrict remedy because “‘any less comprehensive a solution than a metropolitan area plan would result in an all black school system immediately surrounded by practically all white suburban school systems, with an overwhelmingly white majority population in the total metropolitan area.’” Id., at 735. We held that before a district court could order an interdistrict remedy, there must be a showing that “racially discriminatory acts of the state or local school districts, or of a single school district have been a substantial cause of interdistrict segregation.” Id., at 745. Because the record “contained] evidence of de jure segregated conditions only in the Detroit Schools” and there had been “no showing of significant violation by the 53 outlying school districts and no evidence of interdistrict violation or effect,” we reversed the District Court’s grant of interdistrict relief. Ibid.
Justice Stewart provided the Court’s fifth vote and wrote separately to underscore his understanding of the decision. In describing the requirements for imposing an “interdis-trict” remedy, Justice Stewart stated: “Were it to be shown, for example, that state officials had contributed to the separation of the races by drawing or redrawing school district lines; by transfer of school units between districts; or by purposeful, racially discriminatory use of state housing or zoning laws, then a decree calling for the transfer of pupils across district lines or for restructuring of district lines might well be appropriate. In this case, however, no such interdistrict violation was shown.” Id., at 755 (concurring opinion) (citations omitted). Justice Stewart concluded that the Court *94properly rejected the District Court’s interdistrict remedy because “[t]here were no findings that the differing racial composition between schools in the city and in the outlying suburbs was caused by official activity of any sort.” Id., at 757.
What we meant in Milliken I by an interdistrict violation was a violation that caused segregation between adjoining districts. Nothing in Milliken I suggests that the District Court in that case could have circumvented the limits on its remedial authority by requiring the State of Michigan, a constitutional violator, to implement a magnet program designed to achieve the same interdistrict transfer of students that we held was beyond its remedial authority. Here, the District Court has done just that: created a magnet district of the KCMSD in order to serve the mierdistrict goal of attracting nonminority students from the surrounding SSD’s and redistributing them within the KCMSD. The District Court’s pursuit of “desegregative attractiveness” is beyond the scope of its broad remedial authority. See Milliken II, 433 U. S., at 280.
Respondents argue that the District Court’s reliance upon desegregative attractiveness is justified in light of the District Court’s statement that segregation has “led to white flight from the KCMSD to suburban districts.” 1 App. 126; see Brief for Respondents KCMSD et al. 44-45, and n. 28; Brief for Respondents Jenkins et al. 47-49.6 The lower *95courts’ “findings” as to “white flight” are both inconsistent internally,7 and inconsistent with the typical supposition, bolstered here by the record evidence, that “white flight” may result from desegregation, not de jure segregation.8 The United States, as amicus curiae, argues that the District Court’s finding that “de jure segregation in the KCMSD caused white students to leave the system ... is not inconsistent with the district court’s earlier conclusion that the suburban districts did nothing to cause this white flight and therefore could not be included in a mandatory interdistrict remedy.” Brief for United States as Amicus Curiae 19, n. 2; see also post, at 160-164. But the District Court’s earlier findings, affirmed by the Court of Appeals, were not so limited:
“[Contrary to the argument of [plaintiffs] that the [district court] looked only to the culpability of the SSDs, the scope of the order is far broader. ... It noted that only the schools in one district were affected and that the remedy must be limited to that system. In examining the cause and effect issue, the court noted that ‘not only is plaintiff’s evidence here blurred as to cause and *96effect, there is no “careful delineation of the extent of the effect.”’ . . . The district court thus dealt not only with the issue whether the SSDs were constitutional violators but also whether there were significant inter-district segregative effects. . . . When it did so, it made specific findings that negate current significant interdis-trict effects, and concluded that the requirements of Milliken had not been met.” Jenkins v. Missouri, 807 F. 2d 657, 672 (CA8 1986) (affirming, by an equally divided court, the District Court’s findings and conclusion that there was no interdistrict violation or interdistrict effect) (en banc).9
In Freeman, we stated that “[t]he 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.” 503 U. S., at 496. The record here does not support the District Court’s reliance on “white flight” as a justification for a permissible expansion of its intradistrict remedial authority through its pursuit of desegregative attractiveness. See Milliken I, 418 U. S., at 746; see also Dayton Bd. of Ed. v. Brinkman, 433 U. S. 406, 417 (1977) (Dayton I).
Justice Souter claims that our holding effectively overrules Hills v. Gautreaux, 425 U. S. 284 (1976). See also Brief for American Civil Liberties Union et al. as Amici Curiae 18-20. In Gautreaux, the Federal Department of *97Housing and Urban Development (HUD) was found to have participated, along with a local housing agency, in establishing and maintaining a racially segregated public housing program. 425 U. S., at 286-291. After the Court of Appeals ordered “ ‘the adoption of a comprehensive metropolitan area plan,’ ” id., at 291, we granted certiorari to consider the “permissibility in light of [Milliken I] of ‘inter-district relief for discrimination in public housing in the absence of a finding of an inter-district violation.’” Gautreaux, supra, at 292. Because the “relevant geographic area for purposes of the [plaintiffs’] housing options [was] the Chicago housing market, not the Chicago city limits,” 425 U. S., at 299, we concluded that “a metropolitan area remedy ... [was] not impermissible as a matter of law,” id., at 306. Cf. id., at 298, n. 13 (distinguishing Milliken I, in part, because prior cases had established that racial segregation in schools is “to be dealt with in terms of ‘an established geographic and administrative school system’ ”).
In Gautreaux, we did not obligate the District Court to “subjec[t] HUD to measures going beyond the geographical or political boundaries of its violation.” Post, at 171—172. Instead, we cautioned that our holding “should not be interpreted as requiring a metropolitan area order.” Gautreaux, 425 U. S., at 306. We reversed appellate factfinding by the Court of Appeals that would have mandated a metropolitan-area remedy, see id., at 294-295, n. 11, and remanded the case back to the District Court “ ‘for additional evidence and for further consideration of the issue of metropolitan area relief,’ ” id., at 306.
Our decision today is fully consistent with Gautreaux. A district court seeking to remedy an miradistrict violation that has not “directly caused” significant interdistrict effects, Milliken I, supra, at 744-745, exceeds its remedial authority if it orders a remedy with an interdistrict purpose. This conclusion follows directly from Milliken II, decided one year after Gautreaux, where we reaffirmed the bedrock *98principle that “federal-court decrees exceed appropriate limits if they are aimed at eliminating a condition that does not violate the Constitution or does not flow from such a violation.” 433 U. S., at 282. In Milliken II, we also emphasized that “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.” Id., at 280-281. Gautreaux, however, involved the imposition of a remedy upon a federal agency. See 425 U. S., at 292, n. 9. Thus, it did not raise the same federalism concerns that are implicated when a federal court issues a remedial order against a State. See Milliken II, supra, at 280-281.
The District Court’s pursuit of “desegregative attractiveness” cannot be reconciled with our cases placing limitations on a district court’s remedial authority. It is certainly theoretically possible that the greater the expenditure per pupil within the KCMSD, the more likely it is that some unknowable number of nonminority students not presently attending schools in the KCMSD will choose to enroll in those schools. Under this reasoning, however, every increased expenditure, whether it be for teachers, noninstructional employees, books, or buildings, will make the KCMSD in some way more attractive, and thereby perhaps induce nonminority students to enroll in its schools. But this rationale is not susceptible to any objective limitation. Cf. Milliken II, supra, at 280 (remedial decree “must be designed as nearly as possible ‘to restore the victims of discriminatory conduct to the position they would have occupied in the absence of such conduct’ ”). This case provides numerous examples demonstrating the limitless authority of the District Court operating under this rationale. See, e. g., App. to Pet. for Cert. A-115 (The District Court has recognized that it has “provide[d] the KCMSD with facilities and opportunities not available anywhere else in the country”); id., at A-140 (“The District has repeatedly requested that the [District Court] provide ex*99travagant programs based on the hopes that they will succeed in the desegregation effort”). In short, desegregative attractiveness has been used “as the hook on which to hang numerous policy choices about improving the quality of education in general within the KCMSD.” Jenkins II, 495 U. S., at 76 (Kennedy, J., concurring in part and concurring in judgment).
Nor are there limits to the duration of the District Court’s involvement. The expenditures per pupil in the KCMSD currently far exceed those in the neighboring SSD’s. 19 F. 3d, at 399 (Beam, J., dissenting from denial of rehearing en banc) (per-pupil costs within the SSD’s, excluding capital costs, range from $2,854 to $5,956; per-pupil costs within the KCMSD, excluding capital costs, are $9,412); Brief for Respondent KCMSD et al. 18, n. 5 (arguing that per-pupil costs in the KCMSD, excluding capital costs, are $7,665.18). Sixteen years after this litigation began, the District Court recognized that the KCMSD has yet to offer a viable method of financing the “wonderful school system being built.” App. to Pet. for Cert. A-124; cf. Müliken II, supra, at 293 (Powell, J., concurring in judgment) (“Th[e] parties . . . have now joined forces apparently for the purpose of extracting funds from the state treasury”). Each additional program ordered by the District Court — and financed by the State — to increase the “desegregative attractiveness” of the school district makes the KCMSD more and more dependent on additional funding from the State; in turn, the greater the KCMSD’s dependence on state funding, the greater its reliance on continued supervision by the District Court. But our cases recognize that local autonomy of school districts is a vital national tradition, Dayton I, 433 U. S., at 410, and that a district court must strive to restore state and local authorities to the control of a school system operating in compliance with the Constitution. See Freeman, 503 U. S., at 489; Dowell, 498 U. S., at 247.
*100The District Court’s pursuit of the goal of “desegregative attractiveness” results in so many imponderables and is so far removed from the task of eliminating the racial identifi-ability of the schools within the KCMSD that we believe it is beyond the admittedly broad discretion of the District Court. In this posture, we conclude that the District Court’s order of salary increases, which was “grounded in remedying the vestiges of segregation by improving the de-segregative attractiveness of the KCMSD,” App. to Pet. for Cert. Á-90, is simply too far removed from an acceptable implementation of a permissible means to remedy previous legally mandated segregation. See Milliken II, 433 U. S., at 280.
Similar considerations lead us to conclude that the District Court’s order requiring the State to continue to fund the quality education programs because student achievement levels were still “at or below national norms at many grade levels” cannot be sustained. The State does not seek from this Court a declaration of partial unitary status with respect to the quality education programs. Reply Brief for Petitioners 3. It challenges the requirement of indefinite funding of a quality education program until national norms are met, based on the assumption that while a mandate for significant educational improvement, both in teaching and in facilities, may have been justified originally, its indefinite extension is not.
Our review in this respect is needlessly complicated because the District Court made no findings in its order approving continued funding of the quality education programs. See App. to Pet. for Cert. A-69 to A-75. Although the Court of Appeals later recognized that a determination of partial unitary status requires “careful factfinding and detailed articulation of findings,” 11 F. 3d, at 765, it declined to remand to the District Court. Instead it attempted to assemble an adequate record from the District Court’s state*101ments from the bench and subsequent orders. Id., at 761. In one such order relied upon by the Court of Appeals, the District Court stated that the KCMSD had not reached anywhere close to its “maximum potential because the District is still at or below national norms at many grade levels.” App. to Pet. for Cert. A-131.
But this clearly is not the appropriate test to be applied in deciding whether a previously segregated district has achieved partially unitary status. See Freeman, supra, at 491; Dowell, 498 U. S., at 249-250. The basic task of the District Court is to decide whether the reduction in achievement by minority students attributable to prior de jure segregation has been remedied to the extent practicable. Under our precedents, the State and the KCMSD are “entitled to a rather precise statement of [their] obligations under a desegregation decree.” Id., at 246. Although the District Court has determined that “[segregation has caused a system wide reduction in achievement in the schools of the KCMSD,” 639 F. Supp., at 24, it never has identified the incremental effect that segregation has had on minority student achievement or the specific goals of the quality education programs. Cf. Dayton I, supra, at 420.10
In reconsidering this order, the District Court should apply our three-part test from Freeman v. Pitts, supra, at 491. The District Court should consider that the State’s role with respect to the quality education programs has been limited to the funding, not the implementation, of those programs. As all the parties agree that improved achievement on test scores is not necessarily required for the State to achieve partial unitary status as to the quality education programs, the District Court should sharply limit, if not dispense with, its reliance on this factor. Brief for Respond*102ents KCMSD et al. 34-35; Brief for Respondents Jenkins et al. 26. Just as demographic changes independent of de jure segregation will affect the racial composition of student assignments, Freeman, 503 U. S., at 494-495, so too will numerous éxternal factors beyond the control of the KCMSD and the State affect minority student achievement. So long as these external factors are not the result of segregation, they do not figure in the remedial calculus. See Spangler, 427 U. S., at 434; Swann, 402 U. S., at 22. Insistence upon academic goals unrelated to the effects of legal segregation unwarrantably postpones the day when the KCMSD will be able to operate on its own.
The District Court also should consider that many goals of its quality education plan already have been attained: the KCMSD now is equipped with “facilities and opportunities not available anywhere else in the country.” App. to Pet. for Cert. A-115. KCMSD schools received an AAA rating eight years ago, and the present remedial programs have been in place for seven years. See 19 F. 3d, at 401 (Beam, J, dissenting from denial of rehearing en banc). It may be that in education, just as it may be in economics, a “rising tide lifts all boats,” but the remedial quality education program should be tailored to remedy the injuries suffered by the victims of prior de jure segregation. See Milliken II, supra, at 287. Minority students in kindergarten through grade 7 in the KCMSD always have attended AAA-rated schools; minority students in the KCMSD that previously attended schools rated below AAA have since received remedial education programs for a period of up to seven years.
On remand, the District Court must bear in mind that its end purpose is not only “to remedy the violation” to the extent practicable, but also “to restore state and local authorities to the control of a school system that is operating in compliance with the Constitution.” Freeman, supra, at 489.
*103The judgment of the Court of Appeals is reversed.
It is so ordered.
“ ‘Magnet schools,’ as generally understood, are public schools of voluntary enrollment designed to promote integration by drawing students away from their neighborhoods and private schools through distinctive curricula and high quality.” Missouri v. Jenkins, 495 U. S. 33, 40, n. 6 (1990).
In April 1993, 16 years after this litigation began, the District Court acknowledged that the KCMSD and the plaintiffs had “barely addressed ... how the KCMSD proposes to ultimately fund the school system developed under the desegregation plan.” App. to Pet. for Cert. A-123. In the context of a proposal to extend funding of the magnet-school program for 10 additional years at a cost of over $500 million, the District Court noted that “[t]he District’s proposals do not include a viable method of financing any of the programs.” Id., at A-140.
“Whether a federal court order granting salary increases to virtually every employee of a school district — including non-instructional personnel — as part of a school desegregation remedy conflicts with applicable decisions of this court which require that remedial components must directly address and relate to the constitutional violation and be tailored to cure the condition that offends the Constitution?” Pet. for Cert. i.
See also Jenkins v. Missouri, 931 F. 2d 1273, 1274 (CA8 1991) (“[T]he district court in September 1984 held the State defendants and the KCMSD liable for intradistrict segregation”); Jenkins v. Missouri, 931 F. 2d 470, 475 (CA8 1991) (“In a June 5, 1984, order the district court rejected claims of interdistrict violations”); Jenkins v. Missouri, 838 F. 2d 260, 264 (CA8 1988) (“In this case, the plaintiffs made unsuccessful claims against the State as well as the suburban, federal, and Kansas defendants for interdistrict relief. They also made successful intradistrict claims against the State and KCMSD”); Jenkins v. Missouri, 807 F. 2d 657, 669-670 (CA8 1986) (en banc) (“[T]he argument that KCMSD officially sanctioned suburban flight looks first to KCMSD’s violation which the district court clearly found to be only intradistrict in nature”).
See also Green v. School Bd. of New Kent Cty., 391 U. S. 430, 432 (1968) (approving a desegregation plan which had a racial composition of 57% black and 43% white); Wright v. Council of Emporia, 407 U. S. 451, 457 (1972) (approving a desegregation plan which had a racial composition of 66% black and 34% white); United States v. Scotland Neck City Bd. of Ed., 407 U. S. 484, 491, n. 5 (1972) (approving implicitly a desegregation plan which had a racial composition of 77% black and 22% white).
Prior to 1954, Missouri mandated segregated schools for black and white children. Jenkins v. Missouri, 593 F. Supp. 1485, 1490 (WD Mo. 1984). Immediately after the Court’s decision in Brown v. Board of Education, 347 U. S. 483 (1954), the State’s Attorney General issued an opinion declaring the provisions that mandated segregation unenforceable. 593 F. Supp., at 1490. In the 1954-1955 school year, 18.9% of the KCMSD’s students were black. 807 F. 2d, at 680. The KCMSD became 30% black, in the 1961-1962 school year, 40% black in the 1965-1966 school year, and 60% black in the 1975-1976 school year. Ibid. In 1977, the KCMSD implemented the 6C desegregation plan in order to ensure that each school within the KCMSD had a minimum minority enrollment of 30%. Jenkins *95v. Missouri, 639 F. Supp. 19, 35 (WD Mo. 1985). Overall enrollment in KCMSD decreased by 30% from the time that the 6C plan first was implemented until 1986. Id., at 36. During the same time period, white enrollment decreased by 44%. Ibid.
Compare n. 4, supra, and Jenkins, 807 F. 2d, at 662 (“[N]one of the alleged discriminatory actions committed by the State or the federal defendants ha[s] caused any significant current interdistrict segregation”), with Jenkins v. Missouri, 855 F. 2d 1295, 1302 (CA8 1988) (“These holdings are bolstered by the district court’s findings that the preponderance of black students in the district was due to the State and KCMSD’s constitutional violations, which caused white flight”).
“Diming the hearing on the liability issue in this case there was an abundance of evidence that many residents of the KCMSD left the district and moved to the suburbs because of the district’s efforts to integrate its schools.” 1 App. 239; see also Scotland Neck City Bd. of Ed., 407 U. S., at 491 (recognizing that implementation of a desegregation remedy may result in “white flight”).
Justice Souter construes the Court of Appeals’ determination to mean that the violations by the State and the KCMSD did not cause segregation within the limits of each of the SSD’s. Post, at 163-164. But the Court of Appeals would not have decided this question at the behest of these plaintiffs — present and future KCMSD students — who have no standing to challenge segregation within the confines of the SSD’s. Cf. Lujan v. Defenders of Wildlife, 504 U. S. 555, 560-561 (1992). Ergo, the Court of Appeals meant exactly what it said: the requirements of Milliken I had not been met because the District Court’s specific findings “negate current significant interdistrict effects.” Jenkins, 807 F. 2d, at 672.
To the extent that the District Court has adopted the quality education program to further the goal of desegregative attractiveness, that goal is no longer valid. See supra, at 91-100.