dissenting.
This now-federal ease originated as an appeal in state court from a municipal agency’s denials of demolition permits. The review that state law provides is classically appellate in character — on the agency’s record, not de novo. Nevertheless, the Court decides today that this standard brand of appellate review can be shifted from the appropriate state tribunal to a federal court of first instance at the option of either party — plaintiff originally or defendant by removal. The Court approves this enlargement of district court authority explicitly in federal-question eases, and by inescapable implication in diversity cases, satisfied that “neither the jurisdictional statutes nor our prior decisions suggest that federal jurisdiction is lacking.” Ante, at 168.
The Court’s authorization of cross-system appeals qualifies as a watershed decision. After today, litigants asserting federal-question or diversity jurisdiction may routinely lodge in federal courts direct appeals from the actions of all manner of local (county and municipal) agencies, boards, and commissions. Exercising this cross-system appellate authority, federal courts may now directly superintend local agencies by affirming, reversing, or modifying their administrative rulings.
The Court relies on the statutory words found in both 28 U. S. C. §§ 1831 and 1332: “The district courts shall have original jurisdiction of all civil actions ....” Then, as its linchpin, the Court emphasizes the 1990 codification and expansion, in § 1367, of what previously had been known as “ancillary jurisdiction” and “pendent jurisdiction.” Specifically, the Court stresses the broad authorization in § 1367(a) for district court exercise of “supplemental jurisdiction” over claims “so related” to a “civil action of which the district courts have original jurisdiction” as to “form part of the same [Article III] case or controversy.” See ante, at 164-*176169, 171-174.1 The bare words of §§ 1331, 1332, and 1367(a) permit the Court’s construction. For the reasons advanced in this opinion, however, I do not construe these prescriptions, on allocation of judicial business to federal courts of first instance, to embrace the category of appellate business at issue here.
The Court’s expansive reading, in my judgment, takes us far from anything Congress conceivably could have meant. Cf. Lynch v. Overholser, 369 U. S. 705, 710 (1962) (“The decisions of this Court have repeatedly warned against the dangers of an approach to statutory construction which confines itself to the bare words of a statute, for ‘literalness may strangle meaning.’”) (citations omitted). Cross-system appeals, if they are to be introduced into our federal system, should stem from the National Legislature’s considered and explicit decision. In accord with the views of the large majority of federal judges who have considered the question, I would hold the cross-system appeal unauthorized by Congress, and affirm the Seventh Circuit’s judgment.
h-i
Until now it has been taken almost for granted that federal courts of first instance lack authority under §§ 1331 and 1332 to displace state courts as forums for on-the-record review of state and local agency actions. In Chicago, R. I. & P. R. Co. v. Stude, 346 U. S. 574 (1954), we recalled the historic understanding: A federal district court “does not sit to review on appeal action taken administratively or judicially in *177a state proceeding.” Id., at 581.2 Cross-system appellate authority is entrusted to this Court, we said in Rooker v. Fidelity Trust Co., 263 U. S. 413 (1923), but it is outside the domain of the lower federal courts. Interpreting the statutory predecessors of 28 U. S. C. §§ 1331 and 1257, we held in Rooker that a federal district court could not modify a decision of the Indiana Supreme Court, for only this Court could exercise such authority. 263 U. S., at 416.
Today, the Court holds that Congress, by enacting § 1367, has authorized federal district courts to conduct deferential, on-the-record review of local agency decisions whenever a federal question is pended to the agency review action. Dismissing, as irrelevant to jurisdiction, the distinction between de novo and deferential review, the Court also provides easy access to federal court whenever the dissatisfied party in a local agency proceeding has the requisite diverse citizenship. The Court does all this despite the overwhelming weight of lower federal court decisions disclaiming cross-system appellate authority, and without even a hint from Congress that so startling a reallocation of power from state courts to federal courts was within the national lawmakers’ contemplation.3
*178I catalog first the decisions, in addition to the Seventh Circuit’s, that the Court today overrides: Volkswagen de Puerto Rico, Inc. v. Puerto Rico Labor Relations Bd., 454 F. 2d 38, 42 (CA1 1972) (permitting a district court to conduct on-the-record review of a decision of the Puerto Rico Labor Relations Board under § 1331 “would place a federal court in an improper posture vis-a-vis a non-federal agency”); Armistead v. C & M Transport, Inc., 49 F. 3d 43, 47 (CA1 1995) (“As courts of original jurisdiction, federal district courts sitting in diversity jurisdiction do not have appellate power, nor the right to exercise supplementary equitable control over original proceedings in the state’s administrative tribunals.”); Frison v. Franklin County Bd. of Ed., 596 F. 2d 1192, 1194 (CA4 1979) (District Court should have declined pendent jurisdiction over a state-law claim “because it is essentially a petition for judicial review of state administrative action rather than a distinct claim for relief”); Fairfax County Redevelopment & Housing Auth. v. W. M. Schlosser Co., 64 F. 3d 155, 158 (CA4 1995) (“Because the district court is ‘a court of original jurisdiction,’ not ‘an appellate tribunal,’ and, thus, is without jurisdiction ‘to review on appeal action taken administratively or judicially in a state proceeding,’ it was without jurisdiction [under § 1332] to conduct such a review of the County Executive’s finding.”) (citations omitted); Labiche v. Louisiana Patients’ Compensation Fund Oversight Bd., 69 F. 3d 21, 22 (CA5 1995) (“We have reviewed [28 U. S. C. §§ 1330-1368] and none would authorize appellate review by a United States District Court of any actions *179taken by a state agency.”); Shamrock Motors, Inc. v. Ford Motor Co., 120 F. 3d 196, 200 (CA9 1997) (“When a state provides for administrative agency review of an appellate nature, rather than administrative review of a de novo nature, federal district courts have neither original jurisdiction nor removal jurisdiction over the review proceedings.”); Trapp v. Goetz, 373 F. 2d 380, 383 (CA10 1966) (Under § 1332, “a United States District Court could not review an appeal action taken either administratively or judicially in a state proceeding.”). Indeed, research discloses only a single Court of Appeals decision that has approved a federal district court’s exercise of cross-system appellate review. See Range Oil Supply Co. v. Chicago, R. I. & P. R. Co., 248 F. 2d 477, 478-479 (CA8 1957) (District Court could exercise removal jurisdiction over an appeal from a state railroad and warehouse commission once that appeal had been perfected in state court). As the Ninth Circuit said in Shamrock Motors: “[T]he prospect of a federal court sitting as an appellate court over state administrative proceedings is rather jarring and should not be quickly embraced as a matter of policy.” Shamrock Motors, Inc. v. Ford Motor Co., 120 F. 3d, at 200.
Until today, federal habeas corpus proceedings were the closest we had come to cross-system 'appellate review. See 28 U. S. C. §§ 2241-2254.4 Unlike the jurisdictional reallocation the Court now endorses, habeas corpus jurisdiction does not entail direct review of a state or local authority’s decision. See Lambrix v. Singletary, 520 U. S. 518, 523 (1997). *180Notably, in providing for federal habeas corpus review, Congress has taken great care to avoid interrupting or intruding upon state-court processes. See, e. g., 28 U. S. C. § 2254(b)(1) (1994 ed., Supp. III) (requiring exhaustion of state remedies before filing a federal petition for writ of ha-beas corpus). The Court’s holding in this “Chicago” case, however, permits the federal court to supplant the State’s entire scheme for judicial review of local administrative actions.
When a local actor or agency violates a person’s federal right, it is indee'd true that the aggrieved party may bring an action under 42 U. S. C. § 1983 without first exhausting state remedies. See Patsy v. Board of Regents of Fla., 457 U. S. 496, 516 (1982). But such an action involves no disregard, as the cross-system appeal does, of the separateness of state and federal adjudicatory systems. In a § 1983 action, a federal (or state) court inquires whether a person, acting under color of state law, has subjected another “to the deprivation of any rights, privileges, or immunities secured by the Constitution and [federal] laws.” The court exercises original, not appellate, jurisdiction; it proceeds independently, not as substantial evidence reviewer on a nonfederal agency’s record. As now-Chief Judge Posner explained:
“[A] suit under 42 U. S. C. § 1983 is not a review proceeding even when ... it challenges administrative action that has an adjudicative component. Federal courts have no general appellate authority over state courts or state agencies.... The case that is in federal court did not begin in' the state agency but is an independent as well as an original federal action.” Hameetman v. Chicago, 776 F. 2d 636, 640 (CA7 1985).
1 — 1 5 — {
To reach its landmark result, the Court holds that a district court may perform cross-system appellate review of administrative agency decisions so long as the plaintiff’s complaint also contains related federal claims, for “[t]hose federal *181claims suffice to make the actions ‘civil actions’ within the ‘original jurisdiction’ of the district courts.” Ante, at 166. Measuring today’s disposition against prior decisions concerning proceedings in federal court following a state administrative decision, the Court, ante, at 169-171, takes up Horton v. Liberty Mut. Ins. Co., 367 U. S. 348 (1961), and Stude, see supra, at 176-177, and n. 2.
Horton was a workers’ compensation ease proceeding in federal court on the basis of the parties’ diverse citizenship. The contending parties were an injured worker and the insurance company that served as compensation carrier for the worker’s employer. At the administrative stage, the Texas Industrial Accident Board made an award of $1,050. Neither side was satisfied. The insurer maintained that the worker was entitled to no compensation, while the worker urged his entitlement to the statutory maximum of $14,035. The insurer brought suit first, filing its complaint in federal court; one week later, the worker filed a state-court suit and sought dismissal of the insurer’s federal action on alternative grounds: (1) the matter in controversy did not meet § 1332’s monetary amount requirement (then ‘‘in excess of $10,000”); (2) the insurer’s suit was “nothing more than an appeal from a state administrative order” and federal courts have “no appellate jurisdiction” over such orders, 367 U. S., at 354.
After concluding that the jurisdictional amount requirement was met, the Court turned to the question whether the federal-court proceeding was in fact an “appeal,” and therefore barred under Stude, which, as the Horton Court described it, “held that a- United States District Court was without jurisdiction to consider an appeal ‘taken administratively or judicially in a state proceeding.’ ” 367 U. S., at 354 (quoting Stude, 346 U. S., at 581). On that matter, the Texas Supreme Court’s construction of the State’s compensation law left no room for debate. When suit commences, the administrative award is vacated and the court determines liability de novo. See 367 U. S., at 355, n. 15. The suit to set *182aside an award is thus like any other first instance proceeding — it is “‘a suit, not an appeal.’” Id., at 354 (quoting Booth v. Texas Employers’ Ins. Assn., 132 Tex. 237, 246, 123 S. W. 2d 322, 328 (1938)).
Remarkably, the Court today asserts that neither Stude nor Horton “suggested] that jurisdiction turned on whether judicial review of the administrative determination was deferential or de novo.” Ante, at 170; see also ante, at 171 (“The Court [in Horton] did not purport to hold that the de novo standard was a precondition to federal jurisdiction.”). The Court thus easts aside the critical difference between fresh first instance proceedings not tied to a record made by a tribunal lower in the hierarchy, and on-the-record substantial evidence review, which cannot fairly be described as anything but appellate in character.
If, as the Court reasons today, the distinction between de novo and deferential review is inconsequential, then a district court may, indeed must, entertain cross-system, on-the-record appeals from local agency decisions — without regard to the presence or absence of any federal question— whenever the parties meet the diversity-of-eitizenship requirement of § 1332. The Court so confirms by noting that, in accord with Califano v. Sanders, 430 U. S. 99, 105-107 (1977), “district courts routinely conduct deferential review [of federal administrative action] pursuant to their original jurisdiction over federal questions.” Ante, at 171. Just as routinely, it now appears, district courts must “conduct deferential review [of local administrative action] pursuant to their original jurisdiction over [diversity cases].”
The Court’s homogenization of de novo proceedings and appellate review rests on a single ease, Califano v. Sanders. In Sanders, the Court settled a longstanding division of opinion over whether § 10 of the federal Administrative Procedure Act (APA), 5 U. S. C. §§ 701-704, ranked as an independent grant of subject-matter jurisdiction to federal courts, allowing them to review the actions of. federal agencies, *183without regard to the amount in controversy. The Court held that the APA “does not afford an implied grant of subject-matter jurisdiction permitting federal judicial review of agency action.” 480 U. S., at 107. Nevertheless, the Court explained, district court review of federal administrative action — when Congress had not prescribed another review route or specifically excluded review — would persist. Congress had just dropped the amount-in-eontroversy requirement from §1331, thus “fill[ing] the jurisdictional void.” Id, at 106. With the amount-in-controversy deleted, the Court indicated in Sanders, § 1331 would assure fidelity to the presumption that administrative action is subject to judicial review. See id., at 105-106; Abbott Laboratories v. Gardner, 387 U. S. 136, 141 (1967) (courts generally hold agency action nonreviewable “only upon a showing of ‘clear and convincing evidence’ of a contrary legislative intent”); see also Barlow v. Collins, 397 U. S. 159, 166 (1970) (“[JJudicial review of [federal] administrative action is the rule, and nonreviewability an exception which must be demonstrated.”).
Whatever the reason for the rule implicit in Sanders — that federal district courts may engage in on-the-record, substantial evidence review of federal agency actions under § 1331— Chicago homes in on the statutory language. See Brief for Petitioners 11, 30, 39. Section 1331 reads: “The district courts shall have original jurisdiction of all civil actions arising under the Constitution, laws, or treaties of the United States.” If deferential, on-the-record review of a federal agency’s action qualifies as a “civil action” within a district court’s “original jurisdiction,” Chicago urges, then deferential, on-the-record review of local agency action must fit the same bill, i. e., such review must qualify as a “civil action” within the district court’s “original jurisdiction.”
But one of these things is not necessarily like the other. I recognize that the bare and identical words “original jurisdiction” and “civil action” in §§ 1331 and 1332 comport with *184Chicago’s view and that of the Court. See supra, at 175. We would do well, however, to recall in this context a sage and grave warning: “The tendency to assume that a word which appears in two or more legal rules, and so in connection with more than one purpose, has and should have precisely the same scope in all of them, runs all through legal discussions. It has all the tenacity of original sin and must constantly be guarded against.” Cook, “Substance” and “Procedure” in the Conflict of Laws, 42 Yale L. J. 838, 337 (1933).
Cases “arising under the Constitution, laws, or treaties of the United States” within the meaning of § 1331 compose a collection smaller than the one fitting within the similarly worded Clause in Article III of the Constitution, “Cases ... arising under this Constitution, the Laws of the United States, and Treaties made.” See, e.g., Louisville & Nashville R. Co. v. Mottley, 211 U. S. 149, 152 (1908); Shoshone Mining Co. v. Rutter, 177 U. S. 505, 513 (1900). Diversity of citizenship must be complete to proceed under § 1332, see Strawbridge v. Curtiss, 3 Cranch 267, 268 (1806), but it may be “minimal” in interpleader cases brought under § 1335, see State Farm Fire & Casualty Co. v. Tashire, 386 U. S. 523, 530-531 (1967).
Significantly, in assuming available when a person complains about arbitrary federal administrative action, the Court in Sanders never fixed on the words of § 1331, and never even mentioned in relation to that provision the terms “civil action” or “original jurisdiction.” The Court simply concluded from the legislative history that Congress meant to fill “an interstitial gap,” 430 U. S., at 107, i. e., Congress meant to hold federal agencies accountable by making their actions subject to judicial review.
Statutes like the Illinois Administrative Review Law, Ill. Comp. Stat., ch. 735, §§ 5/3-103, 5/3-104 (Supp. 1997), explicitly provide for state-court judicial review of state and local agency decisions. Unlike the federal picture the Court con*185fronted in Sanders, there is no void to fill. The gap to which Sanders attended — the absence of any forum for “nonstatu-tory” review of federal agency decisions unless § 1331 provided one — simply does not exist in a case brought under a state measure like the Illinois Administrative Review Law. I would therefore resist reading Sanders out of context to mandate cross-system appellate review of local agency decisions.
Ill
Just last Term, two Members of today’s majority recognized the vital interest States have in developing and elaborating state administrative law, for that law regulates the citizen’s contact with state and local government at every turn, for example, in gaining life-sustaining public benefits, obtaining a license, or, as in this case, receiving a permit. Last Term’s lead opinion observed:
“In the States there is an ongoing process by which state courts and state agencies work to elaborate an administrative law designed to reflect the State’s own rules and traditions concerning the respective scope of judicial review and administrative discretion. .. . [TJhe elaboration of administrative law ... is one of the primary responsibilities of the state judiciary. Where, as here, the parties invoke federal principles to challenge state administrative action, the courts of the State have a strong interest in integrating those sources of law within their own system for the proper judicial control of state officials.” Idaho v. Coeur d’Alene Tribe of Idaho, 521 U. S. 261, 276 (1997) (principal opinion of Kennedy, I, joined by Rehnquist, C. J.).
Today’s decision jeopardizes the “strong interest” courts of the State have in controlling the actions of local as well as state agencies. State court superintendence can now be displaced or dislodged in any case against a local agency in *186which the parties are of diverse citizenship and in any case in which a Fourteenth Amendment plea can be made.
The Court insists that there is no escape from this erosion of state-court authority. Its explanation is less than compelling. The Court describes as the alternative "ICS’ proposed approach.” See ante, at 167. That approach, according to the Court, would have us determine first “whether [ICS’s] state claims constitute ‘civil actions’ within a district court’s ‘original jurisdiction.’” Ibid. The Court then demolishes the supposed approach by observing that it “would effectively read the supplemental jurisdiction statute out of the bools.” Ibid.; see also ante, at 167-169.
I do not find in ICS’s brief the approach the Court constructs, then destructs. Instead, the argument I do find, see Brief for Respondents 21-24, runs as follows. Chicago has tried to persuade the Court that ICS’s “Complaints for Administrative Review are no different' than civil rights actions.” Id., at 21. See Notice of Removal for Petitioner in No. 91 C 1587 (ND Ill.), App. 15 (“it appears from the face of plaintiffs’ complaint that this is a civil rights complaint”). ICS acknowledged that it might have chosen to bypass on-the-record administrative review in state court, invoking federal jurisdiction under § 1983 instead, without exhausting state remedies. Brief for Respondents 22-24. Had ICS done so, review would have been “plenary in its scope” and would not have been “confined by the administrative record.” Id., at 24. But ICS did not take that path. It proceeded under the Illinois Administrative Review Law seeking resolution of both state-law and federal constitutional issues “in the context of on-the-record administrative review.” Id., at 22. The distinction between the appellate review it sought and the first instance action it did not bring “is crucial,” ICS argued. Ibid.
In sum, from start to finish, ICS sought accurately to portray the Seventh Circuit’s resistance to “federaliz[ing],” without explicit congressional instruction to do so, “garden-*187variety appeals from ... local administrative decisions/’ id., at 3, appeals in which the federal issues ultimately raised “are inextricably intertwined with [the State’s] administrative review scheme,” id., at 4. Not a case in which pendent or supplemental jurisdiction has ever been exercised is touched by the argument ICS in fact made, which trained constantly on the impropriety of eross-system appellate review. Far from urging the Court to “read the supplemental jurisdiction statute out of the books,” ante, at 167, ICS simply asked the Court not to read into § 1367 more than any other tribunal has conceived to be there. What ICS sought to convey, the Court obscures: “[T]he City fail[ed] to cite a single ease in which a federal court specifically assumed pendent or supplemental jurisdiction over an on-the-record state administrative appeal.” Brief for Respondents 24, n. 11.
Even if the Court were correct in maintaining that Congress thrust local administrative agency on-the-record review proceedings into federal court at the option of either party, given diversity or. an ultimate constitutional argument, the Court’s reluctance to “artieulat[e] general standards” for the guidance of the lower courts is puzzling. Cf. Strickland v. Washington, 466 U. S. 668, 698 (1984) (after “artieulat[ing] general standards for judging ineffectiveness [of counsel] claims,” the Court considered it “useful to apply those standards to the facts of th[e] case in order to illustrate the meaning of the general principles”). ICS, seeking such guidance, did not simply “allud[e] to” the District Court’s extraordinary course. Cf. ante, at 174. This is a summary of the points ICS made in urging the impropriety of federal-court retention of the case, assuming, arguendo, federal-court power to keep it. The permits in question were sought under Chicago’s Landmarks Ordinance, a measure “Illinois courts have never had an opportunity to interpret.” Brief for Respondents 4. “The issues of Illinois constitu*188tional law raised by [ICS] have never been decided by Illinois appellate courts.” Ibid. Land-use cases generally, and landmark designations particularly, implicate “local policies” and “local concerns.” Ibid. Yet all this Court is willing to say is that “the District Court properly exercised federal question jurisdiction over the federal claims in ICS’ complaints, and properly recognized that it could thus also exercise supplemental jurisdiction over ICS’ state law claims.” Ante, at 166. The Court’s opinion expresses “no [further] view.” Ante, at 174.
The District Court disposed of tion and due process claims in less than 13 pages of its 63-page opinion, App. to Pet. for Cert. 33a-46a, and then devoted over 40 pages more to the state-law claims. Id., at 46a-89a. That court wrote at greatest length on whether the Landmarks Commission’s conclusions were “Against the Manifest Weight of the Evidence.” Id., at 73a-89a. Finally, the District Court “a£firm[ed] the Commission’s decisions.” Id., at 89a. It would have been in order for this Court to have recalled, in face of the District Court’s federal-claims-first approach, the “fundamental rule of judicial restraint” that federal courts “will not reach constitutional questions in advance of the necessity of deciding them.” Three Affiliated Tribes of Fort Berthold Reservation v. Wold Engineering, P. C., 467 U. S. 138, 157 (1984). As a rule, potentially dispositive state-law challenges, not ultimate constitutional questions, should be cleared first. See, e.g., Hagans v. Lavine, 415 U. S. 528, 546-547 (1974).
When local official actions are contested on state and federal grounds, and particularly when construction of a state measure or local ordinance is at issue, the state questions stand at the threshold. In this case, for example, had ICS’s construction of the Landmarks Ordinance prevailed, no federal constitutional question would have ripened. The Court does note that § 1367(c) “enumerates] the circumstances in which district courts can refuse [to] exercise [supplemental *189jurisdiction],” ante, at 173, but as to that, the Court simply reports: “[T]he District Court decided [judicial economy, convenience, fairness, and comity] would be best served by exercising jurisdiction over ICS’ state law claims,” ibid.5 The Court also mentions, abstractly, that “district courts may be obligated not to decide state law claims (or to stay their adjudication) where one of the abstention doctrines articulated by this Court applies.” Ante, at 174.
Section 1367(e), which concerns supplemental jurisdiction, will have no utility in diversity eases where, if jurisdiction exists, it is generally not within the court’s discretion to “decline.” And lower courts have found our abstention pronouncements “less than pellucid.” See R. Fallon, D. Meltzer, & D. Shapiro, Hart and Wechsler’s The Federal Courts and the Federal System 1247, 1251 (4th ed. 1996). Which of our “various abstention principles,” ante, at 174, should the lower federal courts consult when asked to review as an appellate instance, and affirm, modify, or reverse, a local license or permit denial? To dispel, confesión and advance comity, should the lower courts endeavor to fashion — and will we eventually declare — a “Chicago” abstention doctrine?
Given the state forum to which ICS resorted, and the questions it raised there, see App. 26-35, 76-77, ICS’s primary contention is clear: The Commission should have granted, under state law, demolition permits or an economic hardship exception. I do not comprehend the Court’s reasons for suggesting that the District Court may have acted *190properly in holding on to this case, rather than allowing the state courts to proceed in their normal course.
V
In Ankenbrandt v. Richards, 504 U. S. 689 (1992), we addressed the question whether civil actions for divorce, alimony, or child custody fall within § 1382 when the parties are of diverse citizenship. Nothing in the text of the Constitution or in the words of § 1332 excluded parties from bringing such “civil actions” in federal court. Historically, however, decrees terminating marriages had been considered wholly within the State’s domain. See Barber v. Barber, 21 How. 582 (1859). That understanding, we noted in Anken-brandt, had prevailed “for nearly a century and a half.” 504 U. S., at 694-695. “Given the long passage of time without any expression of congressional dissatisfaction,” we reaffirmed the absence of statutory jurisdiction for federal court adjudication of original civil actions for divorce, alimony, and child custody. Id., at 703. The Court explained that its conclusion was also
“supported by sound policy considerations. . . . [Sjtate courts are more eminently suited to work of this type than are federal courts, which lack the close association with state and local government organizations dedicated to handling [the] issues [involved].” Id., at 703-704.6
*191History and policy tug strongly here as well. There surely has been no “expression of congressional dissatisfaction” with the near-unanimous view of the Circuits that federal courts may not engage in cross-system appellate review, and “the elaboration of [state] administrative law” is a “prim[e] responsibility] of the state judiciary.” Idaho v. Coeur d’Alene Tribe of Idaho, 521 U. S., at 276.
This Court said in Finley v. United States, 490 U. S. 545, 547-548 (1989):
“It remains rudimentary law that ‘[a]s regards all courts of the United States inferior to this tribunal, two things are necessary to create jurisdiction, whether original or appellate. The Constitution must have given to the court the capacity to take it, and an act of Congress must have supplied it.... To the extent that such action is not taken, the power lies dormant.’” (quoting Mayor v. Cooper, 6 Wall. 247, 252 (1868)).
As I see it, no Act of Congress adverts to and authorizes any cross-system appeal from state or local administrative agency to lower federal court. I would await express legislative direction before proceeding down that road. Accordingly, I would affirm the Seventh Circuit’s judgment.
The Court assumes, although § 1367 does not expressly so provide, that the section covers cases originating in a state court and removed to a federal court. Ante, at 165. Although the point has not been briefed, I do not question that assumption. See Steinman, Supplemental Jurisdiction in § 1441 Removed Cases: An Unsurveyed Frontier of Congress’ Handiwork, 35 Ariz. L. Rev. 305, 308-310 (1993) (observing that arguments against application of § 1367 to removed cases “are weak”).
The Court in Stude also made the following statement: “When the proceeding has reached the stage of a perfected appeal and the jurisdiction of the state district court is invoked, it then becomes in its nature a civil action and subject to removal by the defendant to the United States District Court.” Chicago, R. I. & R R. Co. v. Stude, 346 U. S., at 578-579. This statement, made on the way to the Court’s conclusion that the District Court lacked removal jurisdiction, does not carry great weight. It suggests that while the plaintiff in Stude could not have filed the action in federal court initially under § 1332, the defendant could have removed the action to federal court pursuant to § 1441(a). That suggestion is incorrect, for “[o]nly state-court actions that originally could have been filed in federal court may be removed to federal court by the defendant.” Caterpillar Inc. v. Williams, 482 U. S. 386, 392 (1987).
The Court’s holding can embrace the decisions of state, as opposed to local, agencies, only if the State consents to the district court’s jurisdiction. In Pennhurst State School and Hospital v. Halderman, 465 U. S. 89 (1984), *178the Court held it would violate the Eleventh Amendment for a federal court to entertain, without the State’s consent, “a claim that state officials violated state law in carrying out their official responsibilities.” Id., at 121. The Court further held that “this principle applies as well to state-law claims brought into federal court under pendent jurisdiction.” Ibid. Notably, the Court commented in Pennhurst: "[I]t is difficult to think of a greater intrusion on state sovereignty than when a federal court instructs state officials on how to conform their conduct to state law.” Id., at 106.
The Court’s citation to the individuals with Disabilities Education Act (IDEA), ante, at 171, is unpersuasive for two reasons. First, IDEA has its own jurisdictional provision, so it does not concern §§ 1331, 1332, or 1367. See § 615 of the Individuals with Disabilities Education Act Amendments of 1997, Pub. L. 105-17, 111 Stat. 92, to be codified at 20 U. S. C. § 1415(i)(3)(A); Zobrest v. Catalina Foothills School Dist., 509 U. S. 1, 4 (1993). Second, IDEA creates & federal regime. While IDEA may require federal courts to defer to state agency decisions, those decisions are made pursuant to federal legislation.
Ankenbrandt clarified and illustrated “that the domestic relations exception encompasses only cases involving the issuance of a divorce, alimony, or child custody decree”; claims of a kind traditionally adjudicated in federal courts, for example, tort or contract claims, are not excepted from federal-court jurisdiction simply because they arise in a domestic relations context. Ankenbrandt v. Rickards, 504 U. S., at 704. In enacting the Violence Against Women Act of 1994, 108 Stat. 1916, 42 U. S. C. § 13981 et seq., Congress reinforced Ankenbrandt by providing expressly that 28 U. S. C. § 1367 shall not be construed, by reason of a claim arising under the Act, “to confer on the courts of the United States jurisdiction over any State law claim seeking the establishment of a divorce, alimony, equitable distribution of marital property, or child custody decree.” 42 U.S.C. § 13981(e)(4).