concurring in the judgment.
When considering the constitutionality of a state election regulation that restricts core political speech or imposes "severe burdens” on speech or association, we have generally required that the law be narrowly tailored to serve a compelling state interest. But if the law imposes “lesser burdens,” we have said that the State’s important regulatory interests are generally sufficient to justify reasonable, nondiserimina-tory restrictions. The Court today appears to depart from this now-settled approach. In my view, Colorado’s badge, registration, and reporting requirements each must be evaluated -under strict scrutiny. Judged by that exacting standard, I agree with the majority that each of the challenged regulations violates the First and Fourteenth Amendments, and accordingly concur only in the judgment.
States, of course, must regulate their elections to ensure that they are conducted in a fair and orderly fashion. See, e. g., Timmons v. Twin Cities Area New Party, 520 U. S. 351, 358 (1997); Burdick v. Takushi, 504 U. S. 428, 433 (1992). But such regulations often will directly restrict or otherwise burden core political speech and assoeiational rights. . To require that every voting, ballot, and campaign regulation be narrowly tailored to serve a compelling interest "would tie the hands of States seeking to assure that elections are operated equitably and efficiently.” Ibid. Consequently, we have developed (although only recently) a framework for assessing the constitutionality, under the First and Fourteenth Amendments, of state election laws. When a State’s rule imposes severe burdens on speech or association, it must be narrowly tailored to serve a compelling interest; lesser burdens trigger less exacting review, and a State’s important regulatory interests are typically enough to justify reasonable restrictions. Timmons, supra, at 358-359; Burdick, *207supra, at 434; Anderson v. Celebrezze, 460 U. S. 780, 788-790 (1983).
Predictability of decisions in this area is certainly important, but unfortunately there is no bright line separating severe from lesser burdens. When a State’s election law directly regulates core political speech, we have always subjected the challenged restriction to strict scrutiny and required that the legislation be narrowly tailored to serve a compelling governmental interest. See, e.g., Burson v. Freeman, 504 U. S. 191, 198 (1992) (Tennessee law prohibiting solicitation of voters and distribution of campaign literature within 100 feet of the entrance of a polling place); Brown v. Hartlage, 456 U. S. 45, 53-54 (1982) (Kentucky’s regulation of candidate campaign promises); First Nat. Bank of Boston v. Bellotti, 435 U. S. 765, 786 (1978) (Massachusetts law prohibiting certain business entities from making expenditures for the purpose of affecting referendum votes).
Even where a State’s law does not directly regulate core political speech, we have applied strict scrutiny. For example, in Meyer v. Grant, 486 U. S. 414 (1988), we considered a challenge to Colorado’s law making it a felony to pay initiative petition circulators. We applied strict scrutiny because we determined that initiative petition circulation “of necessity involves both the expression of a desire for political change and a discussion of the merits of the proposed change.” Id., at 421. In Citizens Against Rent Control/ Coalition for Fair Housing v. Berkeley, 454 U. S. 290 (1981), we subjected to strict scrutiny a city ordinance limiting contributions to committees formed to oppose ballot initiatives because it impermissibly burdened association and expression. Id., at 294.
When core political speech is at issue, we have ordinarily applied strict scrutiny without first determining that the State’s law severely burdens speech. Indeed, in McIntyre v. Ohio Elections Comm’n, 514 U. S. 334 (1995), the Court suggested that we only resort to our severe/lesser burden *208framework if a challenged election law regulates “the mechanics of the electoral process,” not speech. Id., at 345; but see Eu v. San Francisco County Democratic Central Comm., 489 U. S. 214, 222-223 (1989) (first determining that California’s prohibition on primary endorsements by the official governing bodies of political parties burdened speech and association and then applying strict scrutiny). I suspect that when regulations of core political speech are at issue-it makes little difference whether we determine burden first because restrictions on core political speech so plainly impose a “severe burden.”
When an election law burdens voting and associational interests, our eases are much harder to predict, and I am not at all sure that a coherent distinction between severe and lesser burdens can be culled from them. For example, we have subjected to strict scrutiny Connecticut’s requirement that voters in any party primary be registered members of that party because it burdened the “associational rights of the Party and its members.” Tashjian v. Republican Party of Conn., 479 U. S. 208, 217 (1986). We similarly treated California’s laws dictating the organization and composition of official governing bodies of political parties, limiting the term of office of a party chair, and requiring that the chair rotate between residents of northern and southern California because they “burden[ed] the associational rights of political parties and their members,” Eu, supra, at 231. In Storer v. Brown, 415 U. S. 724 (1974), we applied strict scrutiny to California’s law denying a ballot position to independent candidates who had a registered affiliation with a qualified political party within a year of the preceding primary election, apparently because it “substantially” burdened the rights to vote and associate. Id., at 729, 736.1 And in Nor*209man v. Reed, 502 U. S. 279 (1992), we determined that Illinois’ regulation of the use of party names and its law establishing signature requirements for nominating petitions severely burdened association by limiting new parties’ access to the ballot, and held both challenged laws, as construed by the State Supreme Court, unconstitutional because they were not narrowly tailored. Id., at 288-290, 294. By contrast, we determined that Minnesota’s law preventing a candidate from appearing on the ballot as the choice of more than one party burdened a party’s access to the ballot and its associational rights, but not severely, and upheld the ban under lesser scrutiny. Timmons, 520 U. S., at 368. We likewise upheld Hawaii’s prohibition on write-in voting, which imposed, at most, a “very limited” burden on voters’ freedom of choice and association. Burdick, 504 U. S., at 437.
Colorado argues that its badge, registration, and reporting requirements impose “lesser” burdens, and consequently, each ought to be upheld as serving important state interests. I cannot agree.
A
The challenged badge requirement, Colo. Rev. Stat. §1-40-112(2) (1998), directly regulates the content of speech. The State requires that all petition circulators disclose, at the time they deliver their political message, their names and whether they were paid or unpaid. Therefore, the regulation must be evaluated under strict scrutiny. Moreover, the category of burdened speech is defined by its content— Colorado’s badge requirement does not apply to those who circulate candidate petitions, only to those who circulate initiative or referendum proposals. See generally §1-4-905 (candidate petition circulation requirements). Content-based regulation of speech typically must be narrowly tailored to a compelling state interest. See, e.g., Boos v. *210Barry, 485 U. S. 312, 321 (1988). The State’s dominant justification for its badge requirement is that it helps the public to identify, and the State to apprehend, petition circulators who perpetrate fraud. Even assuming that this is a compelling interest, plainly, this requirement is not narrowly tailored. It burdens all circulators, whether they are responsible for committing fraud or not. In any event, the State has failed to satisfy its burden of demonstrating that fraud is a real, rather than a conjectural, problem. See Turner Broadcasting System, Inc. v. FCC, 512 U. S. 622, 664 (1994); Colorado Republican Federal Campaign Comm. v. Federal Election Comm’n, 518 U. S. 604, 647 (1996) (Thomas, J., concurring in judgment and dissenting in part).2
B
Although Colorado’s registration requirement, Colo. Rev. Stat. § 1-40-112(1) (1998), does not directly regulate speech, it operates in the same fashion that Colorado’s prohibition on paid circulators did in Meyer — the requirement reduces the voices available to convey political messages. We unanimously concluded in Meyer that initiative petition circulation was core political speech. 486 U. S., at 421-422. Colorado’s law making it a felony to pay petition circulators burdened that political expression, we said, because it reduced the number of potential speakers. That reduction limited the size of the audience that initiative proponents and circulators might reach, which in turn made it less likely that initiative proposals would garner the signatures necessary to qualify *211for the ballot. Id., at 422-423. I see no reason to revisit our earlier conclusion. The aim of a petition is to secure political change, and the First Amendment, by way of the Fourteenth Amendment, guards against the State’s efforts to restrict free discussions about matters of public concern.3
Colorado primarily defends its registration requirement on the ground that it ensures that petition circulators are residents, which permits the State to more effectively enforce its election laws against those who violate them.4 The Tenth Circuit assumed, and so do I, that the State has a compelling interest in ensuring that all circulators are residents. Even so, it is clear, as the Court of Appeals decided, that the registration requirement is not narrowly tailored. A large number of Colorado’s residents are not registered voters, as the majority points out, ante, at 193, and the State’s asserted interest could be more precisely achieved through a residency requirement.5
*212c
The District Court and the Court of Appeals both suggested that by forcing proponents to identify paid circulators by name, the reports made it less likely that persons would want to circulate petitions. Therefore, both concluded, the reporting requirement had a chilling effect on core political speech similar to the one we recognized in Meyer. American Constitutional Law Foundation, Inc. v. Meyer, 120 F. 3d 1092, 1096 (CA10 1997); American Constitutional Law Foundation, Inc. v. Meyer, 870 F. Supp. 995, 1003 (Colo. 1994). The District Court- additionally determined that preparation of the required monthly reports was burdensome for and involved additional expense to those supporting an initiative petition. Ibid.
In my view, the burdens that the reporting requirement imposes on circulation are too attenuated to constitute a “severe burden” on core political speech. However, “compelled disclosure, in itself, can seriously infringe on privacy of association and belief guaranteed by the First Amendment.” Buckley v. Valeo, 424 U. S. 1, 64 (1976) (per curiam). In Buckley, because the disclosure requirements of the Federal Election Campaign Act of 1971 encroached on assoeiational rights, we required that they pass a “strict test.” Id., at 66. The same assoeiational interests are burdened by the State’s reporting requirements here, and they must be evaluated under strict scrutiny.
Colorado argues that the “essential purpose” of the reports is to identify circulators. Brief for Petitioner 44. It also claims that its required reports are designed to provide “the press and the voters of Colorado a more complete picture of how money is being spent to get a measure on the ballot.” Ibid. Even assuming that Colorado has a compelling interest in identifying circulators, its law does not serve that interest. The State requires that proponents identify only the names of paid circulators, not all circulators. The interest in requiring a report as to the money paid to each *213circulator by name, as the majority points out, ante, at 201, has not been demonstrated.
The State contends that its asserted interest in providing the press and the electorate with information as to how much money is spent by initiative proponents to advance a particular measure is similar to the governmental interests in providing the electorate with information about how money is spent by a candidate and where it comes from, and in deterring actual corruption and avoiding the appearance of corruption that we recognized in Buckley, supra, at 66-67. However, we have suggested that ballot initiatives and candidate elections involve different considerations. Bellotti, 435 U. S., at 791-792 (“[T]he people in our democracy are entrusted with the responsibility for judging and evaluating the relative merits of conflicting arguments.... [I]f there be any danger that the people cannot evaluate the information and arguments advanced by [one source], it is a danger contemplated by the Framers of the First Amendment”); see also Citizens Against Rent Control, 454 U. S., at 296-298. Indeed, we recognized in Meyer that “the risk of improper conduct... is more remote at the petition stage of an initiative.” 486 U. S., at 427. Similarly, I would think, at the very least, the State’s interest in informing the public of the financial interests behind an initiative proposal is not compelling during the petitioning stage.
As it stands after the lower court decisions, proponents must disclose the amount paid per petition signature and the total amount paid to each circulator, without identifying each circulator, at the time the petition is filed. Monthly disclosures are no longer required.6 Because the respondents did *214not sufficiently brief the question, I am willing to assume, for purposes of this opinion, that Colorado’s interest in having this information made available to the press and its voters — before the initiative is voted upon, but not during circulation — is compelling. The reporting provision as modified by the courts below ensures that the public receives information demonstrating the financial support behind an initiative proposal before voting.
I recognize that in Buckley, although the Court purported to apply strict scrutiny, its formulation of that test was more forgiving than the traditional understanding of that exacting standard. The Court merely required that the disclosure provisions have a “substantial relation,” 424 U. S., at 64, to a “substantial” government interest, id., at 68.7 (The majority appears to dilute Buckley’s formulation even further, stating that Colorado’s reporting requirement must be “substantially related to important governmental interests.” Ante, at 202.) To the extent that Buckley suggests that we should apply a relaxed standard of scrutiny, it is inconsistent with our state election law cases that require the application of traditional strict scrutiny whenever a state law “severely burdens” association, and I would not adhere to it. I would nevertheless decide that the challenged portions of Colorado’s disclosure law are unconstitutional as evaluated under the Buckley standard.
To conclude, I would apply strict scrutiny to each of the challenged restrictions, and would affirm the judgment of *215the Court of Appeals as to eaeh of the three provisions before us. As the majority would apply different reasoning, I concur only in the Court’s judgment.
Although we did not explicitly apply strict scrutiny in Storer, we said that the State’s interest was “not only permissible, but compelling,” and that the device the State chose was “an essential part of its overall mechanism.” 415 U. S., at 736.
The majority is correct to note, ante, at 200, that the Tenth Circuit declined to address whether Colorado’s requirement that the badge disclose whether a circulator is paid or a volunteer, and if paid, the name and telephone number of the payer, would be constitutionally permissible standing alone. Nevertheless, the District Court invalidated §1-40-112(2) in its entirety, American Constitutional Law Foundation, Inc. v. Meyer, 870 F. Supp. 995, 1005 (Colo. 1994), and the Court of Appeals affirmed that decision in full. American Constitutional Law Foundation, Inc. v. Meyer, 120 F. 3d 1092, 1096 (CA10 1997).
There is anecdotal evidence in the briefs that circulators do not discuss the merits of a proposed change by initiative in any great depth. Indeed, National Voter Outreach, Inc., an amicus curiae in support of respondents and, according to its statement of interest, the largest organizer of paid petition circulation drives in the United States, describes most conversations between circulator and prospective petition signer as “brief” Brief for National Voter Outreach, Inc., as Amicus Curiae 21. It gives an example of the typical conversation: ‘“Here, sign this. It will really [tick off California Governor] Pete Wilson.”’ Id., at 21, n. 17. In my view, the level of scrutiny cannot turn on the content or sophistication of a political message. Cf. Colorado Republican Federal Campaign Comm. v. Federal Election Comm'n, 518 U. S. 604, 640 (1996) (“Even a pure message of support, unadorned with reasons, is valuable to the democratic process”). 4
Colorado’s law requires that petition circulators be registered electors, Colo. Rev. Stat. § 1-40-112(1) (1998), and while one must reside in Colorado in order to be a registered voter, § l-2-101(l)(b), Colorado does not have a separate residency requirement for petition circulators at this time. 5
The Court of Appeals did not specifically identify any constitutional problem with the monthly reports to the extent that they require disclosure of proponents’ names and proposed ballot measures for which persons were paid to circulate petitions. But the District Court invalidated the entire monthly reporting requirement, 870 F. Supp., at 1005, and the Court of Appeals affirmed its decision in full. See 120 F. 3d, at 1096.
I have previously noted that the Court in Buckley seemed more forgiving in its review of the contribution provisions than it was with respect to the expenditure rules at issue, even though we purported to strictly scrutinize both. Colorado Republican, 518 U. S., at 640, n. 7 (Thomas, J., concurring in judgment and dissenting in part).