with whom The Chief Justice and Mr. Justice Powell join, dissenting.
The Court’s decision in this case resolves a legal question and is necessarily and properly cast in legal terms. According to the Court, a federal district court, having acquired jurisdiction over a “not wholly insubstantial” federal claim, has power to decide other related claims which lack an independent jurisdictional basis. Applying this analysis to the present case, the Court finds the equal protection claim pleaded by petitioners sufficient to satisfy this somewhat hazy definition of “substan-tiality” and appears to approve the District Court’s exercise of pendent jurisdiction over a claim alleging conflict between state and federal welfare regulations. But since we have been admonished that we may not shut our eyes as judges to what we know as men, the practical as well as the legal consequences of this decision should be squarely faced.
In the wake of King v. Smith, 392 U. S. 309 (1968), and Rosado v. Wyman, 397 U. S. 397 (1970), the lower federal courts have been confronted by a massive influx of cases challenging state welfare regulations. The principal *553claim of plaintiffs in the typical case is that the state regulation conflicts with governing federal regulations and is invalid under the Supremacy Clause of the United States Constitution. This allegation presents a federal claim sufficient to satisfy the first jurisdictional requirement of 28 U. S. C. § 1331,1 the so-called “federal question” jurisdictional statute, but many plaintiffs find the statute’s second requirement, that the matter in controversy exceed the sum of $10,000, impossible to meet. Normally, therefore, these cases would be left, as Congress surely understood when it imposed this jurisdictional limitation, to state courts likewise charged with enforcing the United States Constitution.
To avoid this natural disposition, however, plaintiffs in these cases have turned to 28 U. S. C. § 1343, a more narrowly drawn federal jurisdictional statute requiring no minimum jurisdictional amount. The provision of 28 U. S. C. § 1343 relevant to this case reads:
“The district courts shall have original jurisdiction of any civil action authorized by law to be commenced by any person:
“(3) To redress the deprivation, under color of any State law, statute, ordinance, regulation, custom or usage, of any right, privilege or immunity secured by the Constitution of the United States or by any Act of Congress providing for equal rights of citizens or of all persons within the jurisdiction of the United States . . . .”
*554This Court, however, has never held, and does not hold now, that the Supremacy Clause of the Constitution itself provides a basis for jurisdiction under this section. The Court escapes the need for such a decision by granting the federal courts power to hear the Supremacy Clause claim under a theory of pendent jurisdiction. Finding that plaintiffs here have pleaded an equal protection claim sufficiently substantial to satisfy the requirements of 28 U. S. C. § 1343, the Court seems to suggest that consideration of the Supremacy Clause claim may follow as a matter of course. Since I do not believe that the equal protection claim was sufficient to establish jurisdiction under § 1343, or that the doctrine of pendent jurisdiction was appropriately invoked in this case, I dissent.
I
The history of pendent jurisdiction in this Court is long and complex. Its roots go back to Osborn v. Bank of the United States, 9 Wheat. 738 (1824), where the Court said that the jurisdiction of the federal courts extended not only to federal issues themselves but also to nonfederal issues essential to the settlement of the federal claim. No subsequent decision has cast any doubt upon the wisdom of Mr. Chief Justice Marshall's exposition in that case, since a different result would have forced substantial federal cases into state courts for adjudication simply because they involved nonfederal issues as well as federal ones.2 The doctrine was *555expanded in Siler v. Louisville & Nashville R. Co., 213 U. S. 175 (1909), where the Court upheld the power of a district court, having founded its jurisdiction upon federal constitutional claims, to bypass the constitutional questions and to decide an issue of local law. The Court said that the lower court “had the right to decide all the questions in the case, even though it decided the Federal questions adversely to the party raising them, or even if it omitted to decide them at all, but decided the case on local or state questions only.” 3 But the Court at the same time cautioned: “Of course, the Federal question must not be merely colorable or fraudulently set up for the mere purpose of endeavoring to give the court jurisdiction.” 4
*556The Court returned to the question of pendent jurisdiction in Hurn v. Oursler, 289 U. S. 238 (1933), and Levering & Garrigues Co. v. Morrin, 289 U. S. 103 (1933). The Court in both cases agreed that a substantial federal question was necessary to confer initial jurisdiction on the district court,5 a test that must be met whether or not pendent jurisdiction is involved, and then in Hum further attempted to define the necessary relationship between the pendent claim and the claim conferring jurisdiction. According to the Court, a lower federal court could exercise pendent jurisdiction over a separate *557ground alleged in support of a single cause of action, but not over a separate cause of action itself.6
The Court’s most recent extensive treatment of the subject occurred in Mine Workers v. Gibbs, 383 U. S. 715 (1966). Because Hurn had spoken in terms of “causes of action,” a term which was superseded by the adoption of the Federal Rules of Civil Procedure, Gibbs redefined the necessary relation of the federal and nonfederal claims in more understandable terms. Restating the substan-tiality test in pretty much the language of the earlier cases, the Court then continued:
“The state and federal claims must derive from a common nucleus of operative fact. But if, considered without regard to their federal or state character, a plaintiff’s claims are such that he would ordinarily be expected to try them all in one judicial proceeding, then, assuming substantiality of the federal issues, there is power in federal courts to hear the whole.” Id., at 725 (footnote omitted) (emphasis in original).
This language served to clarify jurisdictional questions which had proved troublesome after Hurn v. Oursler. But, importantly, the decision then went on to emphasize *558that power to hear claims lacking an independent jurisdictional basis should not be exercised indiscriminately. The Court reiterated that “pendent jurisdiction is a doctrine of discretion, not of plaintiff's right/’ id., at 726, and urged that the district courts exercise caution not to abuse that discretion. For example, the Court suggested that
“if the federal claims are dismissed before trial, even though not insubstantial in a jurisdictional sense, the state claims should be dismissed as well.” Ibid, (footnote omitted).
Furthermore, the Court stressed that the relative importance of the claims should be considered:
“Similarly, if it appears that the state issues substantially predominate, whether in terms of proof, of the scope of the issues raised, or of the comprehensiveness of the remedy sought, the state claims may be dismissed without prejudice and left for resolution to state tribunals.” Id., at 726-727.
Although the Court’s language in Gibbs necessarily discussed the- relationship between federal and state claims, much of the opinion’s rationale is applicable when pendent jurisdiction is sought over federal claims lacking an independent jurisdictional basis.7 Of course, a *559decision to deny pendent jurisdiction on the ground that state courts should consider questions of state law naturally involves issues relevant to the question of abstention, a consideration not especially applicable when the pendent claim primarily involves questions of federal law. But the presence of federal questions should not induce federal courts to expand their proper jurisdiction. As previously noted, Congress, by requiring a minimum dollar amount for federal question jurisdiction, made a legislative decision to leave certain claims to state courts. Considerations of convenience and judicial economy may justify hearing those claims when genuine federal business, as contrasted to weak claims .intended merely to secure jurisdiction, is before the federal court, but these considerations should be subordinated to considerations of federalism when the claims without independent jurisdiction constitute “the real body” of the case. In this situation the lower courts should remember that federalism embodies
“a system in which there is sensitivity to the legitimate interests of both State and National Governments, and in which the National Government, anxious though it may be to vindicate and protect federal rights and federal interests, always endeavors to do so in ways that will not unduly interfere with the legitimate activities of the States.” Younger v. Harris, 401 U. S. 37, 44 (1971).
The majority rejects this analysis, seemingly'finding that state courts’ greater familiarity with state law is the only reason for declining pendent jurisdiction under Gibbs. But Congress left to state courts not only those claims involving state law but also those claims involving federal law which it felt did not merit the time of federal courts. This Court now says that federal courts should hear those cases anyway since they can *560render “at least as sure-footed” an interpretation of federal law and are “ ‘particularly appropriate bodies’ ” to do so. This opinion, while it undoubtedly reflects the view of this Court, does not reflect with equal accuracy the purpose of Congress.
In Rosado v. Wyman, 397 U. S. 397 (1970), heavily relied upon by the Court to support its position, there was no intimation that the constitutional claim was a weak one pleaded for the purpose of securing federal jurisdiction over a stronger claim. Rather the constitutional claim proved moot. This Court plainly stated:
“Unlike insubstantiality, which is apparent at the outset, mootness, frequently a matter beyond the control of the parties, may not occur until after substantial time and energy have been expended looking toward the resolution of a dispute that plaintiffs were entitled to bring in a federal court.” Id., at 404.
Thus Rosado does not in any way settle the issue before the Court today. Its holding offers no aid in resolving the real and practical issues that the Court confronts in this case.
The Gibbs decision must be understood in its separate parts. First, the Court held that jurisdiction could not attach unless the claim for which jurisdiction was asserted met the requirement of substantiality and unless the pendent claim was sufficiently related to the jurisdictional claim to constitute a single case under the Constitution. Second, the Court admonished that this jurisdiction, even if found to exist, should be exercised judiciously. The relatively permissive standards applied to the issue of whether the Court could consider a pendent claim were not to guide the ultimate decision of whether the Court should consider the pendent claim. Only where “considerations of judicial economy, conven*561ience and fairness to litigants” were served and where the pendent claim did not predominate in scope or worth over the judicial claim, was the doctrine of pendent jurisdiction to be applied. 383 U. S., at 726. While I am convinced that the District Court lacked jurisdiction over an equal protection claim as thin as this one, even if I am wrong on that point it seems clear to me that its decision to exercise pendent jurisdiction over the Supremacy Clause claim was not based on the discretionary considerations outlined in Gibbs, supra.
II
The District Court simply found the equal protection claim in this case to be “substantial” and proceeded without further discussion to the statutory claim. The Court of Appeals, reversing the determination of the District Court, found the claim to be insubstantial and therefore had no need to go further. This Court merely disagrees on the question of substantiality, reinstating the District Court’s jurisdiction. Unfortunately, this process of analysis seems to me to be wrong both in its treatment of the jurisdictional question and in its failure to treat the discretionary aspects of pendent jurisdiction.
Whatever legal terminology is applied to the equal protection claim of the plaintiffs in this case, the one clear fact is that the claim is not very good. In brief, petitioners, who are recipients of public assistance under the Aid to Families with Dependent Children program, all received funds from New York, over and above their usual monthly' grants, to prevent eviction from their places of lodging for nonpayment of rent. The State, pursuant to a provision of the New York Code of Rules and Regulations challenged in the District Court, sought to recover these unusual, expenditures' by making deductions over the next succeeding months from petitioners’ *562normal monthly grants. In their complaint petitioners contended that the New York recoupment procedure deprived them of equal protection of the laws.8
One searches in vain, either in petitioners’ brief or in the opinions of the District Court or this Court, for any reason why this claim meets even a minimal test of substantiality. It would seem extraordinary if, having paid petitioners more than their normal monthly entitlement in order to meet an emergency situation, the State had not sought to recoup the payments over a period of time. The District Court, finding the claim substantial, cited Bradford v. Juras, 331 F. Supp. 167 (Ore. 1971), a decision by a three-judge district court which found jurisdiction on a similar constitutional claim and then decided the case on statutory grounds. In Bradford, however, the Court simply stated that it had jurisdiction under 28 U. S. C. § 1343 (3) without further discussion.9
The opinion of this Court sheds no more light than did the opinion of the District Court. The Court simply states:
“This reasoning with respect to the rationality of the regulation and its propriety under the Equal Protection Clause may ultimately prove correct, but it is not immediately obvious from the decided cases *563or so 'very plain’ under the Equal Protection Clause.” Ante, at 542.
But cases such as Dandridge v. Williams, 397 U. S. 471 (1970), have largely discredited attacks on legislative decisions about the apportionment of limited state welfare funds. At least where the Court has not found a penalty based on race or considerations such as interstate travel, the legislative judgment is upheld whenever a “conceivable rational basis” exists. Although Dandridge did not “suspend the operation of the Equal Protection Clause” in this area, it assuredly makes this particular claim a marginal one.10
I therefore cannot agree that the equal protection claim pleaded here was sufficient to confer jurisdiction on the District Court. Even assuming that the lower court may refer only to the pleadings in making its determination on the question of jurisdiction, the analysis need not be made, as the majority seems to imply, in a legal vacuum. To say that previous decisions have not foreclosed a question unless a prior case “specifically deal[s]” with the same regulation neglects the second branch of the test enunciated in Levering & Garrigues Co. v. Morrin, 289 U. S. 103 (1933), and repeated in later cases, that a *564claim is insubstantial because “obviously without merit.” Id., at 105. Under today’s rationale it appears sufficient for jurisdiction that a plaintiff is able to plead his claim with a straight face. But a district court should be able to dismiss for want of jurisdiction any claim that plainly carries no hope of success on the merits. This lack of promise in turn could be evident from recent decisions of this Court rejecting claims with a similar thesis or laying down rules which would clearly require dismissal on the merits.
Assuming, however, that the District Court here did have jurisdiction, it seems clear to me that under Gibbs the equal protection claim should not support the Supremacy Clause claim also asserted by petitioners. The test for exercising discretion must be a practical one, involving the type of judgments that a reasonable lawyer, evaluating the respective strengths and weaknesses of his case, might undertake. In this case it is highly improbable that a lawyer familiar with this Court’s cases would place much faith in the success of his equal protection claim. In fact, examination of the complaint itself shows that substantially more attention was paid to the Supremacy Clause claim than to the claims under the Fourteenth Amendment. At the very least, the District Court, before it chose to exercise pendent jurisdiction, should have made an identifiable determination that the Equal Protection Clause was not simply asserted for the purpose of giving the Court jurisdiction over the heart of the plaintiffs’ case. To my mind this seems to be a classic case of the statutory tail wagging the constitutional dog.
Ill
Thus, even if the Court of Appeals may have erroneously resolved the question of jurisdiction, the result it reached was correct in terms of the wise exercise of jurisdiction. Whether the equal protection claim pleaded in *565this case meets the threshold of substantiality for jurisdiction in the federal courts, the claim surely should not convince a district court that its main purpose was anything other than to secure jurisdiction for the more promising Supremacy Clause claim. Presented with this situation, the District Court should have declined to exercise pendent jurisdiction over the Supremacy Clause claim and referred the equal protection claim to a three-judge court.11 Since its failure to do so seems to me an abuse of discretion under Gibbs, I dissent.
The relevant provision of 28 U. S. C. §1331 reads as follows: “(a) The district courts shall have original jurisdiction of all civil actions wherein the matter in controversy exceeds the sum or value of $10,000, exclusive of interest and costs, and arises under the Constitution, laws, or treaties of the United States."
The jurisdictional amount was raised from $3,000 to $10,000 in 1958.
“Under this construction, the judicial power of the Union extends effectively and beneficially to that most important class of cases, which depend on the character of the cause. On the opposite construction, the judicial power never can be extended to a whole case, as expressed by the constitution, but to those parts of cases only which present the particular question involving the construction of *555the constitution or the law. We say it never can be extended to the whole case, because, if the circumstance that other points are involved in it, shall disable Congress from authorizing the Courts of the Union to take jurisdiction of the original cause, it equally disables Congress from authorizing those Courts to take jurisdiction of the whole cause, on an appeal, and thus will be restricted to a single question in that cause; and words obvious& intended to secure to those who claim rights under the constitution, laws, or treaties of the United States, a trial in the federal Courts, will be restricted to the insecure remedy of an appeal upon an insulated point, after it has received that, shape which may be given to it by another tribunal, into which he is forced against his will.” Osborn v. Bank of the United States, 9 Wheat. 738, 822-823 (1824).
213 U. S., at 191.
Id., at 191-192. In Siler the Court specifically noted that the constitutional claim was not fraudulently pleaded to confer jurisdiction over the pendent claim.
The Court today, by its heavy emphasis on deciding state issues in preference to constitutional ones, ante, at 546-547, seems to imply that this doctrine should be controlling even when a constitutional claim is pleaded “for the mere purpose of endeavoring to give the court jurisdiction.” I cannot agree. The numerous cases cited in *556the Court’s opinion stand for the long-recognized and sensible policy that cases should be decided on nonconstitutional grounds where possible; but they do not stand for the proposition that claims which would be otherwise dismissed under the principles discussed in Mine Workers v. Gibbs, 383 U. S. 715 (1966), should be heard simply to avoid the constitutional claim which conferred jurisdiction in the first place. See n. 11, infra. In such cases the competing and equally important policy of safeguarding the limited jurisdiction of the federal courts is entitled to more weight than the Court appears to give it.
The Court in Levering, supra, stated:
“Whether an objection that a bill or a complaint fails to state a case under a federal statute raises a question of jurisdiction or of merits is to be determined by the application of a well settled rule. If the bill or the complaint sets forth a substantial claim, a. case is presented within the federal jurisdiction, however the court, upon consideration, may decide as to the legal sufficiency of the facts alleged to support the claim. But jurisdiction, as distinguished from merits, is wanting where the claim set forth in the pleading is plainly unsubstantial. The cases have stated the rule in a variety of ways, but all to that effect. . . . And the federal question averred may be plainly unsubstantial either because obviously without merit, or 'because its unsoundness so clearly results from the previous decisions of this court as to foreclose the subject and leave no room for the inference that the questions sought to be raised can be the subject of controversy.’ ” 289 U. S., at 105-106.
Hurn v. Oursler, 289 U. S. 238, 245-246 (1933):
“But the rule does not go so far as to permit a federal court to assume jurisdiction of a separate and distinct non-federal cause of action because it is joined in the same complaint with a federal cause of action. The distinction to be observed is between a case where two distinct grounds in support of a single cause of action are alleged, one only of which presents a federal question, and a case where two separate and distinct causes of action are alleged, one only of which is federal in character. In the former, where the federal question averred is not plainly wanting in substance, the federal court, even though the federal ground be not established, may nevertheless retain and dispose of the case upon the non-federal ground; in the latter it may not do so upon the non-federal cause of action.” (Emphasis in original.)
The Court in Mine Workers v. Gibbs, 383 U. S., at 727, also stated:
“[Recognition of a federal court's wide latitude to decide ancillary questions of state law does not imply that it must tolerate a litigant’s effort to impose upon it what is in effect only a state law case. Once it appears that a state claim constitutes the real body of a case, to which the federal claim is only an appendage, the state claim may fairly be dismissed.”
I also see no reason why federal courts should be required to “tolerate” efforts to impose upon them federal cases which Congress has chosen to leave to the state courts.
The portion of the petitioners’ complaint setting forth their equal protection claim states in full:
“Said regulation irrationally and invidiously discriminates against plaintiff victims of eviction. No basis exists in law or fact, consistent with the purposes of the Social Security Act, for reducing the level of payments to plaintiffs who are then forced to live far below the subsistence levels provided to all other persons. Said regulation applies a wholty different standard in determining the grant levels of plaintiffs than the income 'resource and exemptions from levy standard, applicable to all other persons in violation of the Equal Protection Clause of the Fourteenth Amendment to the United States Constitution.”
331 F. Supp., at 168.
The Court in Dandridge stated:
“Conflicting claims of morality and intelligence are raised by opponents and proponents of almost every measure, certainly including the one before us. But the intractable economic, social, and even philosophical problems presented by public welfare assistance programs are not the business of this Court. The Constitution may impose certain procedural safeguards upon systems of welfare administration, Goldberg v. Kelly, [397 U. S. 254 (1970)]. But the Constitution does not empower this Court to second-guess state officials charged with the difficult responsibility of allocating limited public welfare funds among the myriad of potential recipients. Cf. Steward Mach. Co. v. Davis, 301 U. S. 548, 584-585; Helvering v. Davis, 301 U. S. 619, 644.” 397 U. S., at 487.
Petitioners originally sought to convene a three-judge court to consider their constitutional claims but later withdrew that request. Pursuant to a stipulation between the parties, the case was then tried before a single judge on the issue of the claimed statutory conflict only. Goosby v. Osser, 409 U. S. 512 (1973), specifies that a three-judge court must be convened to hear constitutional questions within its jurisdiction if they are “substantial.” It is true, of course, that federal courts commonly avoid deciding constitutional questions when alternative grounds for decision are available. See, e. g., Ashwander v. TV A, 297 U. S. 288, 346-347 (1936) (Brandeis, J., concurring). But application of that principle to cases in which the constitutional claim is pleaded primarily to confer jurisdiction over a pendent claim would lead to circular reasoning. Under that theory a claim for which Congress provided no jurisdiction and which a single judge determined to be improperly brought into federal court would become a preferred ground for decision simply because the court wished to avoid the claim over which Congress granted jurisdiction in the first place. To turn to the pendent claim when pendent jurisdiction is properly assumed under Gibbs may be appropriate, but the presence of a constitutional claim which might therefore be avoided should not itself be an independent basis for hearing the pendent claim.
In rare cases, of course, a three-judge court may disagree with the single judge's view that a constitutional claim lacks merit and resolve the constitutional issue in the plaintiff’s favor. At that point, the plaintiff will have his relief, and the case need go no further. Concededly, a constitutional decision will have been rendered when a statutory decision might have been possible, but that cost, in the few cases where it is likely to arise, seems less expensive than the cost of allowing federal jurisdiction to be unnecessarily expanded.