delivered the opinion of the Court.
The concept of a federal general common law, lurking (to use Justice Holmes’ phrase) as a “brooding omnipresence in the sky,” was questioned for some time before being firmly rejected in Erie R. Co. v. Tompkins, 304 U. S. 64 (1938). See Southern Pacific Co. v. Jensen, 244 U. S. 205, 222 (1917) (Holmes, J., dissenting); Black & White Taxicab & Transfer Co. v. Brown & Yellow Taxicab & Transfer Co., 276 U. S. 518, 533 (1928) (dissenting opinion). Erie mandates that a federal court sitting in diversity apply the substantive law of the forum State, absent a federal statutory or constitutional directive to the contrary. 304 U. S., at 78. See also 28 U. S. C. §1652 (“The laws of the several states, except where the Constitution or treaties of the United States or Acts of Congress otherwise require or provide, shall be regarded as rules of decision in civil actions in the courts of the *227United States, in cases where they apply”). In decisions after Erie, this Court made clear that state law is to be determined in the same manner as a federal court resolves an evolving issue of federal law: “with the aid of such light as [is] afforded by the materials for decision at hand, and in accordance with the applicable principles for determining state law.” Meredith v. Winter Haven, 320 U. S. 228, 238 (1943). See also Ruhlin v. New York Life Ins. Co., 304 U. S. 202, 208-209 (1938) (“Application of the ‘State law’ to the present case . . . does not present the disputants with duties difficult or strange”).
In this case, we must decide specifically whether a federal court of appeals may review a district court’s determination of state law under a standard less probing than that applied to a determination of federal law.
r — I
The issue presented arises out of a contract dispute between a college and one of its students. Petitioner Salve Regina College is an institution of higher education located in Newport, R. I. Respondent Sharon L. Russell was admitted to the college and began her studies as a freshman in 1982. The following year, respondent sought admission to the college’s nursing department in order to pursue a bachelor of science degree in nursing. She was accepted by the department and began her nursing studies in the fall of 1983.
Respondent, who was 5'6" tall, weighed in excess of 300 pounds when she was accepted in the nursing program. Immediately after the 1983 school year began, respondent’s weight became a topic of commentary and concern by officials of the nursing program. Respondent’s first year in the program was marked by a series of confrontations and negotiations concerning her obesity and its effect upon her ability to complete the clinical requirements safely and satisfactorily. During her junior year, respondent signed a document that was designated as a “contract” and conditioned her further *228participation in the nursing program upon weekly attendance at a weight-loss seminar and a realized average loss of two pounds per week. When respondent failed to meet these commitments, she was asked to withdraw from the program and did so. She transferred to a nursing program at another college, but had to repeat her junior year in order to satisfy the transferee institution’s 2-year residency requirement. As a consequence, respondent’s nursing education took five years rather than four. She also underwent surgery for her obesity. In 1987, respondent successfully completed her nursing education, and she is now a registered nurse.
Soon after leaving Salve Regina College, respondent filed this civil action in the United States District Court for the District of Rhode Island. She asserted, among others, claims based on (1) intentional infliction of emotional distress, (2) invasion of privacy, and (8) nonperformance by the college of its implied agreement to educate respondent.1 Subject-matter jurisdiction in the District Court was based on diversity of citizenship. See 28 U. S. C. §1332. The parties agree that the law of Rhode Island applies to all substantive aspects of the action. See Erie R. Co. v. Tompkins, supra.
At the close of plaintiff-respondent’s case in chief, the District Court directed a verdict for the individual defendants on all three of the remaining claims, and for the college on the claims for intentional infliction of emotional distress and *229invasion of privacy. App. 82. The court, however, denied the college’s motion for a directed verdict on the breach-of-contract claim, reasoning that “a legitimate factual issue” remained concerning whether “there was substantial performance by the plaintiff in her overall contractual relationship at Salve Regina.” Id., at 88.
At the close of all the evidence, the college renewed its motion for a directed verdict. It argued that under Rhode Island law the strict commercial doctrine of substantial performance did not apply in the general academic context. Therefore, according to petitioner, because respondent admitted she had not fulfilled the terms of the contract, the college was entitled to judgment as a matter of law.
The District Court denied petitioner’s motion. Id., at 92. Acknowledging that the Supreme Court of Rhode Island, to that point, had limited the application of the substantial-performance doctrine to construction contracts, the District Court nonetheless concluded, as a matter of law, that the Supreme Court of Rhode Island would apply that doctrine to the facts of respondent’s case. Id., at 90-91. The Federal District Judge based this conclusion, in part, on his observation that “I was a state trial judge for 18 and !4 years, and I have a feel for what the Rhode Island Supreme Court will do or won’t do.” Id., at 91. Accordingly, the District Court submitted the breach-of-contract claim to the jury. The court instructed the jury:
“The law provides that substantial and not exact performance accompanied by good faith is what is required in a ease of a contract of this type. It is not necessary that the plaintiff have fully and completely performed every item specified in the contract between the parties. It is sufficient if there has been substantial performance, not necessarily full performance, so long as the substantial performance was in good faith and in compliance with the contract, except for some minor and relatively unimportant deviation or omission.” Id., at 97.
*230The jury returned a verdict for respondent, and determined that the damages were $30,513.40. Id., at 113. Judgment was entered. Id., at 115. Both respondent and petitioner appealed.
The United States Court of Appeals for the First Circuit affirmed. 890 F. 2d 484 (1989). It first upheld the District Court’s directed verdict dismissing respondent’s claims for intentional infliction of emotional distress and invasion of privacy. Id., at 487-488. It then turned to petitioner’s argument that the District Court erred in submitting the breach-of-contract claim to the jury. Rejecting petitioner’s argument that, under Rhode Island law, the doctrine of substantial performance does not apply in the college-student context, the court stated:
“In this case of first impression, the district court held that the Rhode Island Supreme Court would apply the substantial performance standard to the contract in question. In view of the customary appellate deference accorded to interpretations of state law made by federal judges of that state, Dennis v. Rhode Island Hospital Trust Nat’l Bank, 744 F. 2d 893, 896 (1st Cir. 1984); O’Rourke v. Eastern Air Lines Inc., 730 F. 2d 842, 847 (2d Cir. 1984), we hold that the district court’s determination that the Rhode Island Supreme Court would apply standard contract principles is not reversible error.” Id., at 489.
Petitioner college sought a writ of certiorari from this Court. It alleged that the Court of Appeals erred in deferring to the District Court’s determination of state law.2 A *231majority of the Courts of Appeals, although varying in their phraseology, embrace a rule of deference similar to that articulated by the Court of Appeals in this case. See, e. g., Norton v. St. Paul Fire & Marine Ins. Co., 902 F. 2d 1355, 1357 (CA8 1990) (“In general, we accord substantial deference to a district court’s interpretation of the law of the state in which it sits”), and Self v. Wal-Mart Stores, Inc., 885 F. 2d 336, 339 (CA6 1989) (“[W]e should give ‘considerable weight’ to the trial court’s views on such questions of local law”). Two Courts of Appeals, however, have broken ranks recently with their sister Circuits. They have concluded that a district-court determination of state law is subject to plenary review by the appellate court. See Craig v. Lake Asbestos of Quebec, Ltd., 843 F. 2d 145, 148 (CA3 1988), and In re McLinn, 739 F. 2d 1395 (CA9 1984) (en banc, with a divided vote). We granted certiorari to resolve the conflict. 497 U. S. 1023 (1990).
II
We conclude that a court of appeals should review de novo a district court’s determination of state law. As a general matter, of course, the courts of appeals are vested with plenary appellate authority over final decisions of district courts. See 28 U. S. C. § 1291. The obligation of responsible appellate jurisdiction implies the requisite authority to review independently a lower court’s determinations.
Independent appellate review of legal issues best serves the dual goals of doctrinal coherence and economy of judicial administration. District judges preside alone over fast-paced trials: Of necessity they devote much of their energy and resources to hearing witnesses and reviewing evidence. Similarly, the logistical burdens of trial advocacy limit the extent to which trial counsel is able to supplement the district judge’s legal research with memoranda and briefs. Thus, *232trial judges often must resolve complicated legal questions without benefit of “extended reflection [or] extensive information.” Coenen, To Defer or Not to Defer: a Study of Federal Circuit Court Deference to District Court Rulings on State Law, 73 Minn. L. Rev. 899, 923 (1989).
Courts of appeals, on the other hand, are structurally suited to the collaborative juridical process that promotes de-cisional accuracy. With the record having been constructed below and settled for purposes of the appeal, appellate judges are able to devote their primary attention to legal issues. As questions of law become the focus of appellate review, it can be expected that the parties’ briefs will be refined to bring to bear on the legal issues more information and more comprehensive analysis than was provided for the district judge. Perhaps most important, courts of appeals employ multijudge panels, see 28 U. S. C. §§ 46(b) and (e), that permit reflective dialogue and collective judgment. Over 30 years ago, Justice Frankfurter accurately observed:
“Without adequate study there cannot be adequate reflection; without adequate reflection there cannot be adequate discussion; without adequate discussion there cannot be that fruitful interchange of minds which is indispensable to thoughtful, unhurried decision and its formulation in learned and impressive opinions.” Dick v. New York Life Ins. Co., 359 U. S. 437, 458-459 (1959) (dissenting opinion).
Independent appellate review necessarily entails a careful consideration of the district court’s legal analysis, and an efficient and sensitive appellate court at least will naturally consider this analysis in undertaking its review. Petitioner readily acknowledges the importance of a district court’s reasoning to the appellate court’s review. See Tr. of Oral Arg. 11,19-22. Any expertise possessed by the district court will inform the structure and content of its conclusions of law and thereby become evident to the reviewing court. If the court of appeals finds that the district court’s analytical sophistica*233tion and research have exhausted the state-law inquiry, little more need be said in the appellate opinion. Independent review, however, does not admit of unreflective reliance on a lower court’s inarticulable intuitions. Thus, an appropriately respectful application of de novo review should encourage a district court to explicate with care the basis for its legal conclusions. See Fed. Rule Civ. Proc. 52(a) (requiring the district court to “state separately its conclusions of law”).
Those circumstances in which Congress or this Court has articulated a standard of deference for appellate review of district-court determinations reflect an accommodation of the respective institutional advantages of trial and appellate courts. In deference to the unchallenged superiority of the district court’s factfinding ability, Rule 52(a) commands that a trial court’s findings of fact “shall not be set aside unless clearly erroneous, and due regard shall be given to the opportunity of the trial court to judge of the credibility of the witnesses.” In addition, it is “especially common” for issues involving supervision of litigation to be reviewed for abuse of discretion. See Pierce v. Underwood, 487 U. S. 552, 558, n. 1 (1988). Finally, we have held that deferential review of mixed questions of law and fact is warranted when it appears that the district court is “better positioned” than the appellate court to decide the issue in question or that probing appellate scrutiny will not contribute to the clarity of legal doctrine. Miller v. Fenton, 474 U. S. 104, 114 (1985); see also Cooter & Gell v. Hartmarx Corp., 496 U. S. 384, 402 (1990) (“[T]he district court is better situated than the court of appeals to marshal the pertinent facts and apply the fact-dependent legal standard mandated by Rule 11”); Pierce, 487 U. S., at 562 (“[T]he question whether the Government’s litigating position has been ‘substantially justified’ is . . . a multifarious and novel question, little susceptible, for the time being at least, of useful generalization”).
Nothing about the exercise of diversity jurisdiction alters these functional components of decisionmaking or otherwise *234warrants departure from a rule of independent appellate review. Actually, appellate deference to the district court’s determination of state law is inconsistent with the principles underlying this Court’s decision in Erie. The twin aims of the Erie doctrine — “discouragement of forum-shopping and avoidance of inequitable administration of the laws,” Hanna v. Plumer, 380 U. S. 460, 468 (1965)—are components of the goal of doctrinal coherence advanced by independent appellate review. As respondent has conceded, deferential appellate review invites divergent development of state law among the federal trial courts even within a single State. Tr. of Oral Arg. 34-36. Moreover, by denying a litigant access to meaningful review of state-law claims, appellate courts that defer to the district courts’ state-law determinations create a dual system of enforcement of state-created rights, in which the substantive rule applied to a dispute may depend on the choice of forum. Cf. Erie, 304 U. S., at 74-75 (“[The rule of Swift v. Tyson, 16 Pet. 1 (1842)] made rights enjoyed under the unwritten ‘general law’ vary according to whether enforcement was sought in the state or in the federal court”). Neither of these results, unavoidable in the absence of independent appellate review, can be reconciled with the commands of Erie.
Although some might say that this Court has not spoken with a uniformly clear voice on the issue of deference to a district judge’s determination of state law, a careful consideration of our cases makes apparent the duty of appellate courts to provide meaningful review of such a determination. In a series of cases decided soon after Erie, the Court noted that the appellate courts had applied general federal law instead of the law of the respective States, and remanded to the Courts of Appeals for consideration of the applicable principles of state law. See, e. g., New York Life Ins. Co. v. Jackson, 304 U. S. 261 (1938), and Rosenthal v. New York Life Ins. Co., 304 U. S. 263 (1938). It is true that in Bernhardt v. Polygraphic Co. of America, 350 U. S. 198 (1956), *235this Court remanded the case to the District Court for application of state law. The Court noted, however, that the law of the State was firmly settled, and emphasized: “Were the question in doubt or deserving further canvass, we would of course remand the case to the Court of Appeals to pass on this question of [state] law.” Id., at 205.3
III
In urging this Court to adopt the deferential standard embraced by the majority of the Courts of Appeals, respondent offers two arguments. First, respondent suggests that the appellate courts professing adherence to the rule of deference actually are reviewing de novo the district-court determinations of state law. Second, respondent presses the familiar contention that district judges are better arbiters of unsettled state law because they have exposure to the judicial system of the State in which they sit. We reject each of these arguments.
A
Respondent primarily contends that the Courts of Appeals that claim to accord special consideration to the District Court’s state-law expertise actually undertake plenary review of a determination of state law. According to respondent, this is simply de novo review “cloth[ed] in ‘deferential’ robes.” Brief for Respondent 15. In support of this contention, respondent refers to several decisions in which the appellate court has announced that it is bound to review deferentially a district court’s determination of state law, yet nonetheless has found that determination to constitute re*236versible error. Afram Export Corp. v. Metallurgiki Halyps, S. A., 772 F. 2d 1358, 1370 (CA7 1985); Norton v. St. Paul Fire & Marine Ins. Co., 902 F. 2d 1355 (CA8 1990). Respondent also relies on eases in which the Courts of Appeals, while articulating a rule of deference, acknowledge their obligation to scrutinize closely the District Court’s legal conclusions. See Foster v. National Union Fire Ins. Co. of Pittsburgh, 902 F. 2d 1316 (CA8 1990). See also In re McLinn, 739 F. 2d, at 1405 (dissenting opinion) (“The majority overreacts to a problem that is basically one of terminology”).
We decline the invitation to assume that courts of appeals craft their opinions disingenuously. The fact that an appellate court overturns an erroneous determination of state law in no way indicates that the appellate court is not applying the rule of deference articulated in the opinion. The cases cited by respondent confirm this. In Foster, for example, the Court of Appeals articulated a rule of deference, yet cautioned: “We have not, however, failed to closely examine the matter ourselves.” 902 F. 2d, at 1318. Respondent would have us interpret this caveat as an acknowledgment of the appellate court’s obligation to review the state-law question de novo. See Brief for Respondent 17-18, and n. 23. The Court of Appeals, however, expressly acknowledged that it would not reverse the District Court’s determination “unless its analysis is ‘fundamentally deficient . . . , without a reasonable basis, or contrary to a reported state-court opinion.’” Foster, 902 F. 2d, at 1318 (citations omitted). After reviewing the applicable law in some detail, the Court of Appeals concluded: “[T]he district court’s interpretation of the applicable Arkansas law is certainly not deficient in analysis and is reasonable.” Id., at 1320. This neither purports to be, nor is, a conclusion following from de novo review.
Nor does it suffice to recognize that little substantive difference may separate the form of deference articulated and applied by the several Courts of Appeals and the independent appellate review urged by petitioner. Respondent argues *237that the subtle differences between these standards are insufficient to warrant intrusion into the manner in which appellate courts review state-law determinations. A variation of this argument forms the framework upon which the dissent in McLinn rests. See 739 F. 2d, at 1404 (“By giving ‘substantial deference,’ or. . . ‘great weight,’ to the decisions of the district courts, appellate courts do not suspend their own thought processes”).
As a practical matter, respondent and the dissent in McLinn frequently may be correct. We do not doubt that in many cases the application of a rule of deference in lieu of independent review will not affect the outcome of an appeal. In many diversity cases the controlling issues of state law will have been squarely resolved by the state courts, and a district court’s adherence to the settled rule will be indisputably correct. See, e. g., Bernhardt, 350 U. S., at 204-205. In a case where the controlling question of state law remains unsettled, it is not unreasonable to assume that the considered judgment of the court of appeals frequently will coincide with the reasoned determination of the district court. Where the state-law determinations of the two courts diverge, the choice between these standards of review is of no significance, if the appellate court concludes that the district court was clearly wrong.4
Thus, the mandate of independent review will alter the appellate outcome only in those few cases where the appellate *238court would resolve an unsettled issue of state law differently from the district court’s resolution, but cannot conclude that the district court’s determination constitutes clear error. See, e. g., In re McLinn, 739 F. 2d, at 1397 (“The panel indicated that if the question of law were reviewed under the deferential standard that we have applied in the past, which permits reversal only for clear error, then they would affirm; but if they were to review the determination under an independent de novo standard, they would reverse”). These few instances, however, make firm our conviction that the difference between a rule of deference and the duty to exercise independent review is “much more than a mere matter of degree.” Bose Corp. v. Consumers Union of United States, Inc., 466 U. S. 485, 501 (1984). When de novo review is compelled, no form of appellate deference is acceptable.
B
Respondent and her amicus also argue that de novo review is inappropriate because, as a general matter, a district judge is better positioned to determine an issue of state law than are the judges on the court of appeals. This superior capacity derives, it is said, from the regularity with which a district judge tries a diversity case governed by the law of the forum State, and from the extensive experience that the district judge generally has had as practitioner or judge in the forum State. See Brief for Respondent 7-10; Brief for Ford Motor Co. as Amicus Curiae 9-11.
We are unpersuaded. As an initial matter, this argument seems to us to be founded fatally on overbroad generalizations. Moreover, and more important, the proposition that a district judge is better able to “intuit” the answer to an unsettled question of state law is foreclosed by our holding in Erie. The very essence of the Erie doctrine is that the bases of state law are presumed to be communicable by the parties to a federal judge no less than to a state judge. Almost 35 years ago, Professor Kurland stated: “Certainly, if the law *239is not a brooding omnipresence in the sky over the United States, neither is it a brooding omnipresence in the sky of Vermont, or New York or California.” Kurland, Mr. Justice Frankfurter, the Supreme Court and the Erie Doctrine in Diversity Cases, 67 Yale L. J. 187, 217 (1957). See Southern Pacific Co., 244 U. S., at 222 (Holmes, J., dissenting) (“The common law is not a brooding omnipresence in the sky but the articulate voice of some sovereign or quasi-sovereign that can be identified”). Similarly, the bases of state law are as equally communicable to the appellate judges as they are to the district judge. To the extent that the available state law on a controlling issue is so unsettled as to admit of no reasoned divination, we can see no sense in which a district judge’s prior exposure or nonexposure to the state judiciary can be said to facilitate the rule of reason.5
>
The obligation of responsible appellate review and the principles of a cooperative judicial federalism underlying Erie require that courts of appeals review the state-law determinations of district courts de novo. The Court of Appeals in this *240case therefore erred in deferring to the local expertise of the District Court.
The judgment of the Court of Appeals is reversed, and the case is remanded for further proceedings consistent with this opinion.
It is so ordered.
The amended complaint named the college and five faculty members as defendants and alleged discrimination in violation of the Rehabilitation Act of 1973, 87 Stat. 355, as amended, 29 U. S. C. § 701 et seq.; denial of due process and unconstitutional interference with her liberty and property interests; negligent and intentional infliction of emotional distress; invasion of privacy; wrongful dismissal; violation of express and implied covenants of good faith and fair dealing; and breach of contract. The District Court entered summary judgment for the defendants except as to the three state-law claims for intentional infliction of emotional distress, invasion of privacy, and breach of contract. 649 F. Supp. 391, 407 (1986). It determined that it need not consider “the plausibility of federal question jurisdiction.” Id., at 393, n. 1.
See Coenen, To Defer or Not to Defer: A Study of Federal Circuit Court Deference to District Court Rulings on State Law, 73 Minn. L. Rev. 899 (1989), and the many cases cited therein. See also Note, What is the Proper Standard for Reviewing a District Court’s Interpretation of State Substantive Law?, 54 U. Cin. L. Rev. 215 (1985), and Note, A Nondeferen-tial Standard for Appellate Review of State Law Decisions by Federal District Courts, 42 Wash. & Lee L. Rev. 1311 (1985). See, however, Woods, *231The Erie Enigma: Appellate Review of Conclusions of Law, 26 Ariz. L. Rev. 755 (1984), and Note, The Law/Fact Distinction and Unsettled State Law in the Federal Courts, 64 Texas L. Rev. 157 (1985).
The dissent inexplicably relies on several cases in which this Court declined to review de novo questions of state law to support the dissent’s contention that it is “quite natural” for appellate judges to rely on the “experience” of district judges. See post, at 241-242. We are not persuaded that the manner in which this Court chooses to expend its limited resources in the exercise of its discretionary jurisdiction has any relevance to the obligation of courts of appeals to review de novo those legal issues properly before them.
Of course, a question of state law usually can be resolved definitively if the litigation is instituted in state court and is not finally removed to federal court, or if a certification procedure is available and is successfully utilized. Rhode Island provides a certification procedure. See Rhode Island Supreme Court Rule 6 (1989).
See, however, Lehman Brothers v. Schein, 416 U. S. 386, 390-391 (1974) (“We do not suggest that where there is doubt as to local law and where the certification procedure is available, resort to it is obligatory. It does, of course, in the long run save time, energy, and resources and helps build a cooperative judicial federalism. Its use in a given case rests in the sound discretion of the federal court”) (footnote omitted).
“As a general proposition, a federal court judge who sits in a particular state, especially one who has practiced before its courts, may be better able to resolve complex questions as to the law of that state than is a federal judge who has no such personal acquaintance with the law of the state. For this reason federal appellate courts frequently have voiced reluctance to substitute their own view of the state law for that of the district judge. As a matter of judicial administration, this seems defensible. But there is some tendency to go beyond that proposition and to say that if the trial court has reached a permissible conclusion under state law, the appellate court cannot reverse even if it thinks the state law to be otherwise, thereby treating the question of state law much as if it were a question of fact. The determination of state law, however, is a legal question, and although the considered decision of a district judge experienced in the law of a state naturally commands the respect of an appellate court, a party is entitled to meaningful review of that decision just as he is of any other legal question in the case, and just as he would have been if the case had been tried in a state court.” 19 C. Wright, A. Miller, & E. Cooper, Federal Practice and Procedure § 4507, pp. 106-110 (1982) (footnotes omitted).