concurring in part and concurring in the judgment.
This is not a hard case. If there is any difficulty, it arises from the problem of obtaining a consensus on the meaning of the word “policy” — a word that does not appear in the text of 42 U. S. C. § 1983, the statutory provision that we are supposed to be construing. The difficulty is thus a consequence of this Court’s lawmaking efforts rather than the work of the Congress of the United States.1
With respect to both the merits of the constitutional claim and the county’s liability for the unconstitutional activities of its agents performed in the course of their official duties, there can be no doubt that the Congress that enacted the Ku Klux Act in 1871 intended the statute to authorize a recovery in a case of this kind. When police officers chopped down the door to petitioner’s premises in order to serve capiases on two witnesses, they violated petitioner’s constitutional rights. Steagald v. United States, 451 U. S. 204 (1981), *488makes it perfectly clear that forcible entry to a third party’s premises to execute an arrest warrant is unconstitutional if the entry is without a search warrant and in the absence of consent or exigent circumstances.2 In my view, it is not at all surprising that respondents have “conceded” the retro-activity of Steagald. For Steagald plainly presented its holding as compelled by, and presaged in, well-established precedent.3
*489Similarly, if we view the question of municipal liability from the perspective of the Legislature that enacted the Ku Klux Act of 1871, the answer is clear. The legislative history indicating that Congress did not intend to impose civil liability on municipalities for the conduct of third parties, ante, at 478-479, and n. 7, merely confirms the view that it did intend to impose liability for the governments’ own illegal acts — including those acts performed by their agents in the course of their employment. In other words, as I explained in my dissent in Oklahoma City v. Tuttle, 471 U. S. 808, 835-840 (1985), both the broad remedial purpose of the statute and the fact that it embodied contemporaneous common-law doctrine, including respondeat superior, require a conclusion that Congress intended that a governmental entity be hable for the constitutional deprivations committed by its agents in the course of their duties.4
*490Finally, in construing the scope of § 1983, the Court has sometimes referred to “considerations of public policy.”5 To the extent that such “policy” concerns are relevant, they also support a finding of county liability. A contrary construction would produce a most anomalous result. The primary responsibility for protecting the constitutional rights of the residents of Hamilton County from the officers of Hamilton County should rest on the shoulders of the county itself, rather than on the several agents who were trying to perform their jobs. Although I recognize that the county may provide insurance protection for its agents, I believe that the primary party against whom the judgment should run is the county itself. The county has the resources and the authority that can best avoid future constitutional violations and provide a fair remedy for those that have occurred in the past. Thus, even if “public policy” concerns should inform the construction of § 1983, those considerations, like the statute’s remedial purpose and common-law background, support *491a conclusion of county liability for the unconstitutional, axe-swinging entry in this case.
Because I believe that Parts I, II-A, and II-C are consistent with the purpose and policy of § 1983, as well as with our precedents, I join those Parts of the Court’s opinion6 and concur in the judgment.
See Oklahoma City v. Tuttle, 471 U. S. 808, 841 (1985) (Stevens, J., dissenting) (“While the Court purports to answer a question of statutory construction ... its opinion actually provides us with an interpretation of the word ‘policy’ as it is used in Part II of the opinion in Monell v. New York City Dept. of Social Services, 436 U. S. 658, 690-695 (1978). The word ‘policy’ does not appear in the text of § 1983, but it provides the theme for today’s decision”). It may be significant that the issue has apparently become, not the purpose and scope of 42 U. S. C. § 1983, but the nature of the liability “envisioned” by this Court “in Monell.” Post, at 491 (O’Connor, J., concurring in part and concurring in judgment); post, at 499 (Powell, J., dissenting).
Indeed, it can be argued that the justification for a forcible entry to serve a capias, as in this case, is even weaker than the justification for a forcible entry to execute an arrest warrant, as in Steagald. Since the Sixth Circuit in this action, 746 F. 2d 337 (1984), and the Ohio Supreme Court in reviewing petitioner’s conviction, State v. Pembaur, 9 Ohio St. 3d 136, 459 N. E. 2d 217, cert. denied, 467 U. S. 1219 (1984), did not distinguish between the two situations, however, and since the forcible entry was unconstitutional under either conception, it is unnecessary to rest on that possible difference.
See 451 U. S., at 211 (“Except in [cases of consent or exigent circumstances], we have consistently held that the entry into a home to conduct a search or make an arrest is unreasonable under the Fourth Amendment unless done pursuant to a warrant”); id., at 213-214 (“In the absence of exigent circumstances, we have consistently held that such judicially untested determinations are not reliable enough to justify an entry into a person’s home to arrest him without a warrant, or a search of a home for objects in the absence of a search warrant. . . . We see no reason to depart from this settled course when the search of a home is for a person rather than an object”); id., at 216 (“Since warrantless searches of a home are impermissible absent consent or exigent circumstances, we conclude that the instant search violated the Fourth Amendment”); id., at 219 (“[I]f anything, the little guidance that can be gleaned from common-law authorities undercuts the Government’s position. The language of Semayne’s Case, [5 Co. Rep. 91a, 77 Eng. Rep. 194 (K. B. 1603)], . . . suggests that although the subject of an arrest warrant could not find sanctuary in the home of the third party, the home remained a ‘castle or privilege’ for its residents. Similarly several [common-law] commentators suggested that a search warrant, rather than an arrest warrant, was necessary to fully insulate a constable from an action for trespass brought by a party whose home was searched”); id., at 220 (“[T]he history of the Fourth Amendment strongly suggests that its Framers would not have sanctioned the instant search”).
The fact that the Sixth Circuit and two other Circuits had reached a contrary conclusion does not transform Steagald into a nonretroactive opinion. This Court has never suggested that resolution of a split in the Circuits *489somehow means that a holding is presumptively nonretroactive in the Circuits that have disagreed with the Court’s ultimate conclusion. Furthermore, the suggestion that there is a more compelling need for nonretroactivity in a civil context than in a criminal context, post, at 492-496 (Powell, J., dissenting), ignores the fact that, in a civil context, there is not the societal cost of reversing convictions. Cf. Johnson v. New Jersey, 384 U. S. 719, 731 (1966) (“[R]etroactive application of Escobedo [v. Illinois, 378 U. S. 478 (1964),] and Miranda [v. Arizona, 384 U. S. 436 (1966),] would seriously disrupt the administration of our criminal laws. It would require the retrial or release of numerous prisoners found guilty by trustworthy evidence in conformity with previously announced constitutional standards”). Additionally, Payton v. New York, 445 U. S. 573 (1980), which Steagald cites and discusses, has, of course, been held retroactive in the only context in which the Court has considered the issue. See United States v. Johnson, 457 U. S. 537 (1982).
Several commentators have concluded that the dicta in Monell v. New York City Dept. of Social Services, 436 U. S. 658 (1978), regarding respondeat superior misreads the legislative history of § 1983. See, e. g., Blum, From Monroe to Monell: Defining the Scope of Municipal Liability in Federal Courts, 51 Temp. L.Q. 409, 413, n. 15 (1978) (“The interpretation adopted by the Court with respect to the rejection of vicarious liability under § 1983 had been espoused prior to Monell by one author who drew a distinction between ‘political’ § 1983 cases, in which a city itself causes the constitutional violation, and ‘constitutional tort’ § 1983 cases, in which an *490attempt is made to impose vicarious liability on the city for the misconduct of its employees. . . . Although this view of § 1983 may represent a sensitive response to the fiscal plight of municipal corporations today, it should not be acknowledged as a legitimate interpretation of congressional intent in 1871”); Note, Section 1983 Municipal Liability and the Doctrine of Respondeat Superior, 46 U. Chi. L. Rev. 935, 936 (1979) (“[T]he purposes and legislative history of the provision demand a scheme of respondeat superior liability”); Note, Monell v. Department of Social Services: One Step Forward and a Half Step Back for Municipal Liability Under Section 1983, 7 Hofstra L. Rev. 893, 921 (1979) (“Analysis of the legislative history of section 1983 does not indicate that Congress intended to exclude respondeat superior from the act. The language of the statute similarly offers no such proof. Since both were relied on by the Court in Monell, the dicta in that decision is, at best, poorly reasoned authority for the proposition that a municipality is not liable for the unauthorized acts of its employees”); Comment, Municipal Liability under Section 1983 for Civil Rights Violations After Monell, 64 Iowa L. Rev. 1032, 1045 (1979) (“The Court’s [respondeat superior] limitation ... is not justified by the legislative history of section 1983 or by policy considerations”).
Newport v. Fact Concerts, Inc., 453 U. S. 247, 266 (1981); Owen v. City of Independence, 445 U. S. 622, 650 (1980).
The reasons for my not joining Parts II and IV of Monell, 436 U. S. at 714 (Stevens, J., concurring in part), are also applicable to my decision not to join Part II-B of this opinion.