Respondent filed this suit under 42 U. S. C. § 1983 alleging that petitioner violated the Fourteenth Amendment's Due Process Clause when its police officers, acting pursuant to official policy or custom, failed to respond to her repeated reports over several hours that her estranged husband had taken their three children in violation of her restraining order against him. Ultimately, the husband murdered the children. The District Court granted the town's motion to dismiss, but an en banc majority of the Tenth Circuit reversed, finding that respondent had alleged a cognizable procedural due process claim because a Colorado statute established the state legislature's clear intent to require police to enforce restraining orders, and thus its intent that the order's recipient have an entitlement to its enforcement. The court therefore ruled, among other things, that respondent had a protected property interest in the enforcement of her restraining order.

Held: Respondent did not, for Due Process Clause purposes, have a property interest in police enforcement of the restraining order against her husband. Pp. 755-768.

(a) The Due Process Clause's procedural component does not protect everything that might be described as a government "benefit": "To have a property interest in a benefit, a person . . . must . . . have a legitimate claim of entitlement to it." Board of Regents of State Colleges v. Roth, 408 U. S. 564, 577. Such entitlements are created by existing rules or understandings stemming from an independent source such as state law. E.g., ibid. Pp. 755-756.

(b) A benefit is not a protected entitlement if officials have discretion to grant or deny it. See, e. g., Kentucky Deft, of Corrections v. Thompson, 490 U. S. 454, 462-463. It is inappropriate here to defer to the Tenth Circuit's determination that Colorado law gave respondent a right to police enforcement of the restraining order. This Court therefore proceeds to its own analysis. Pp. 756-758.

(c) Colorado law has not created a personal entitlement to enforcement of restraining orders. It does not appear that state law truly made such enforcement mandatory. A well-established tradition of police discretion has long coexisted with apparently mandatory arrest statutes. Cf. Chicago v. Morales, 527 U. S. 41, 47, n. 2, 62, n. 32. Against that backdrop, a true mandate of police action would require some stronger indication than the Colorado statute's direction to "use every reasonable means to enforce a restraining order" or even to "arrest . . . or . . . seek a warrant." A Colorado officer would likely have some discretion to determine that — despite probable cause to believe a restraining order has been violated — the violation's circumstances or competing duties counsel decisively against enforcement in a particular instance. The practical necessity for discretion is particularly apparent in a case such as this, where the suspected violator is not actually present and his whereabouts are unknown. In such circumstances, the statute does not appear to require officers to arrest but only to seek a warrant. That, however, would be an entitlement to nothing but procedure, which cannot be the basis for a property interest. Pp. 758-764.

(d) Even if the statute could be said to make enforcement "mandatory," that would not necessarily mean that respondent has an entitlement to enforcement. Her alleged interest stems not from common law or contract, but only from a State's statutory scheme. If she was given a statutory entitlement, the Court would expect to see some indication of that in the statute itself. Although the statute spoke of "protected person[s]" such as respondent, it did so in connection with matters other than a right to enforcement. Most importantly, it spoke directly to the protected person's power to "initiate" contempt proceedings if the order was issued in a civil action, which contrasts tellingly with its conferral of a power merely to "request" initiation of criminal contempt proceedings — and even more dramatically with its complete silence about any power to "request" (much less demand) that an arrest be made. Pp. 764-766.

(e) Even were the Court to think otherwise about Colorado's creation of an entitlement, it is not clear that an individual entitlement to enforcement of a restraining order could constitute a "property" interest for due process purposes. Such a right would have no ascertainable monetary value and would arise incidentally, not out of some new species of government benefit or service, but out of a function that government actors have always performed — arresting people when they have probable cause. A benefit's indirect nature was fatal to a due process claim in O'Bannon v. Town Court Nursing Center, 447 U. S. 773, 787. Here, as there, "[t]he simple distinction between government action that directly affects a citizen's legal rights . . . and action that is directed against a third party and affects the citizen only . . . incidentally, provides a sufficient answer to" cases finding government-provided services to be entitlements. Id., at 788. Pp. 766-768.

SCALIA, J., delivered the opinion of the Court, in which REHNQUIST, C. J., and O'CONNOR, KENNEDY, SOUTER, THOMAS, and BREYER, JJ., joined. SOUTER, J., filed a concurring opinion, in which BREYER, J., joined, post, p. 769. STEVENS, J., filed a dissenting opinion, in which GINSBURG, J., joined, post, p. 773.

JUSTICE SCALIA delivered the opinion of the Court.

We decide in this case whether an individual who has obtained a state-law restraining order has a constitutionally protected property interest in having the police enforce the restraining order when they have probable cause to believe it has been violated.

The horrible facts of this case are contained in the complaint that respondent Jessica Gonzales filed in Federal District Court. (Because the case comes to us on appeal from a dismissal of the complaint, we assume its allegations are true. See Swierkiewicz v. Sorema N. A., 534 U. S. 506, 508, n. 1 (2002).) Respondent alleges that petitioner, the town of Castle Rock, Colorado, violated the Due Process Clause of the Fourteenth Amendment to the United States Constitution when its police officers, acting pursuant to official policy or custom, failed to respond properly to her repeated reports that her estranged husband was violating the terms of a restraining order.

The restraining order had been issued by a state trial court several weeks earlier in conjunction with respondent's divorce proceedings. The original form order, issued on May 21, 1999, and served on respondent's husband on June 4, 1999, commanded him not to "molest or disturb the peace of [respondent] or of any child," and to remain at least 100 yards from the family home at all times. 366 F. 3d 1093, 1143 (CA10 2004) (en banc) (appendix to dissenting opinion of O'Brien, J.). The bottom of the preprinted form noted that the reverse side contained "IMPORTANT NOTICES FOR RESTRAINED PARTIES AND LAW ENFORCEMENT OFFICIALS." Ibid, (emphasis deleted). The pre-printed text on the back of the form included the following "WARNING":

"A KNOWING VIOLATION OF A RESTRAINING ORDER IS A CRIME. . . . A VIOLATION WILL ALSO CONSTITUTE CONTEMPT OF COURT. YOU MAY BE ARRESTED WITHOUT NOTICE IF A LAW ENFORCEMENT OFFICER HAS PROBABLE CAUSE TO BELIEVE THAT YOU HAVE KNOWINGLY VIOLATED THIS ORDER." Id., at 1144 (emphasis in original).

The preprinted text on the back of the form also included a "NOTICE TO LAW ENFORCEMENT OFFICIALS," which read in part:

"YOU SHALL USE EVERY REASONABLE MEANS TO ENFORCE THIS RESTRAINING ORDER. YOU SHALL ARREST, OR, IF AN ARREST WOULD BE IMPRACTICAL UNDER THE CIRCUMSTANCES, SEEK A WARRANT FOR THE ARREST OF THE RESTRAINED PERSON WHEN YOU HAVE INFORMATION AMOUNTING TO PROBABLE CAUSE THAT THE RESTRAINED PERSON HAS VIOLATED OR ATTEMPTED TO VIOLATE ANY PROVISION OF THIS ORDER AND THE RESTRAINED PERSON HAS BEEN PROPERLY SERVED WITH A COPY OF THIS ORDER OR HAS RECEIVED ACTUAL NOTICE OF THE EXISTENCE OF THIS ORDER." Ibid. (same).

On June 4, 1999, the state trial court modified the terms of the restraining order and made it permanent. The modified order gave respondent's husband the right to spend time with his three daughters (ages 10, 9, and 7) on alternate weekends, for two weeks during the summer, and, "`upon reasonable notice,'" for a midweek dinner visit "`arranged by the parties'"; the modified order also allowed him to visit the home to collect the children for such "parenting time." Id., at 1097 (majority opinion).

According to the complaint, at about 5 or 5:30 p.m. on Tuesday, June 22, 1999, respondent's husband took the three daughters while they were playing outside the family home. No advance arrangements had been made for him to see the daughters that evening. When respondent noticed the children were missing, she suspected her husband had taken them. At about 7:30 p.m., she called the Castle Rock Police Department, which dispatched two officers. The complaint continues: "When [the officers] arrived . . ., she showed them a copy of the TRO and requested that it be enforced and the three children be returned to her immediately. [The officers] stated that there was nothing they could do about the TRO and suggested that [respondent] call the Police Department again if the three children did not return home by 10:00 p.m." App. to Pet. for Cert. 126a.

At approximately 8:30 p.m., respondent talked to her husband on his cellular telephone. He told her "he had the three children [at an] amusement park in Denver." Ibid. She called the police again and asked them to "have someone check for" her husband or his vehicle at the amusement park and "put out an [all points bulletin]" for her husband, but the officer with whom she spoke "refused to do so," again telling her to "wait until 10:00 p.m. and see if" her husband returned the girls. Id., at 126a-127a.

At approximately 10:10 p.m., respondent called the police and said her children were still missing, but she was now told to wait until midnight. She called at midnight and told the dispatcher her children were still missing. She went to her husband's apartment and, finding nobody there, called the police at 12:10 a.m.; she was told to wait for an officer to arrive. When none came, she went to the police station at 12:50 a.m. and submitted an incident report. The officer who took the report "made no reasonable effort to enforce the TRO or locate the three children. Instead, he went to dinner." Id., at 127a.

At approximately 3:20 a.m., respondent's husband arrived at the police station and opened fire with a semiautomatic handgun he had purchased earlier that evening. Police shot back, killing him. Inside the cab of his pickup truck, they found the bodies of all three daughters, whom he had already murdered. Ibid.

On the basis of the foregoing factual allegations, respondent brought an action under Rev. Stat. § 1979, 42 U. S. C. § 1983, claiming that the town violated the Due Process Clause because its police department had "an official policy or custom of failing to respond properly to complaints of restraining order violations" and "tolerate[d] the non-enforcement of restraining orders by its police officers." App. to Pet. for Cert. 129a. The complaint also alleged that the town's actions "were taken either willfully, recklessly or with such gross negligence as to indicate wanton disregard and deliberate indifference to" respondent's civil rights. Ibid.

Before answering the complaint, the defendants filed a motion to dismiss under Federal Rule of Civil Procedure 12(b)(6). The District Court granted the motion, concluding that, whether construed as making a substantive due process or procedural due process claim, respondent's complaint failed to state a claim upon which relief could be granted.

A panel of the Court of Appeals affirmed the rejection of a substantive due process claim, but found that respondent had alleged a cognizable procedural due process claim. 307 F. 3d 1258 (CA10 2002). On rehearing en banc, a divided court reached the same disposition, concluding that respondent had a "protected property interest in the enforcement of the terms of her restraining order" and that the town had deprived her of due process because "the police never `heard' nor seriously entertained her request to enforce and protect her interests in the restraining order." 366 F. 3d, at 1101, 1117. We granted certiorari. 543 U. S. 955 (2004).

The Fourteenth Amendment to the United States Constitution provides that a State shall not "deprive any person of life, liberty, or property, without due process of law." Amdt. 14, § 1. In 42 U. S. C. § 1983, Congress has created a federal cause of action for "the deprivation of any rights, privileges, or immunities secured by the Constitution and laws." Respondent claims the benefit of this provision on the ground that she had a property interest in police enforcement of the restraining order against her husband; and that the town deprived her of this property without due process by having a policy that tolerated nonenforcement of restraining orders.

As the Court of Appeals recognized, we left a similar question unanswered in DeShaney v. Winnebago County Dept. of Social Servs., 489 U. S. 189 (1989), another case with "undeniably tragic" facts: Local child-protection officials had failed to protect a young boy from beatings by his father that left him severely brain damaged. Id., at 191-193. We held that the so-called "substantive" component of the Due Process Clause does not "requir[e] the State to protect the life, liberty, and property of its citizens against invasion by private actors." Id., at 195. We noted, however, that the petitioner had not properly preserved the argument that — and we thus "decline[d] to consider" whether — state "child protection statutes gave [him] an `entitlement' to receive protective services in accordance with the terms of the statute, an entitlement which would enjoy due process protection." Id., at 195, n. 2.

The procedural component of the Due Process Clause does not protect everything that might be described as a "benefit": "To have a property interest in a benefit, a person clearly must have more than an abstract need or desire" and "more than a unilateral expectation of it. He must, instead, have a legitimate claim of entitlement to it." Board of Regents of State Colleges v. Roth, 408 U. S. 564, 577 (1972). Such entitlements are, "`of course, . . . not created by the Constitution. Rather, they are created and their dimensions are defined by existing rules or understandings that stem from an independent source such as state law.'" Paul v. Davis, 424 U. S. 693, 709 (1976) (quoting Roth, supra, at 577); see also Phillips v. Washington Legal Foundation, 524 U. S. 156, 164 (1998).

Our cases recognize that a benefit is not a protected entitlement if government officials may grant or deny it in their discretion. See, e.g., Kentucky Dept. of Corrections v. Thompson, 490 U. S. 454, 462-463 (1989). The Court of Appeals in this case determined that Colorado law created an entitlement to enforcement of the restraining order because the "court-issued restraining order . . . specifically dictated that its terms must be enforced" and a "state statute command[ed]" enforcement of the order when certain objective conditions were met (probable cause to believe that the order had been violated and that the object of the order had received notice of its existence). 366 F. 3d, at 1101, n. 5; see also id., at 1100, n. 4; id., at 1104-1105, and n. 9. Respondent contends that we are obliged "to give deference to the Tenth Circuit's analysis of Colorado law on" whether she had an entitlement to enforcement of the restraining order. Tr. of Oral Arg. 52.

We will not, of course, defer to the Tenth Circuit on the ultimate issue: whether what Colorado law has given respondent constitutes a property interest for purposes of the Fourteenth Amendment. That determination, despite its state-law underpinnings, is ultimately one of federal constitutional law. "Although the underlying substantive interest is created by `an independent source such as state law/ federal constitutional law determines whether that interest rises to the level of a `legitimate claim of entitlement' protected by the Due Process Clause." Memphis Light, Gas Water Div. v. Craft, 436 U. S. 1, 9 (1978) (quoting Roth, supra, at 577; (emphasis added); cf. United States ex rel. TVA v. Powelson, 319 U. S. 266, 279 (1943). Resolution of the federal issue begins, however, with a determination of what it is that state law provides. In the context of the present case, the central state-law question is whether Colorado law gave respondent a right to police enforcement of the restraining order. It is on this point that respondent's call for deference to the Tenth Circuit is relevant.

We have said that a "presumption of deference [is] given the views of a federal court as to the law of a State within its jurisdiction." Phillips, supra, at 167. That presumption can be overcome, however, see Leavitt v. Jane L., 518 U. S. 137, 145 (1996) (per curiam), and we think deference inappropriate here. The Tenth Circuit's opinion, which reversed the Colorado District Judge, did not draw upon a deep well of state-specific expertise, but consisted primarily of quoting language from the restraining order, the statutory text, and a state-legislative-hearing transcript. See 366 F. 3d, at 1103-1109. These texts, moreover, say nothing distinctive to Colorado, but use mandatory language that (as we shall discuss) appears in many state and federal statutes. As for case law: The only state-law cases about restraining orders that the Court of Appeals relied upon were decisions of Federal District Courts in Ohio and Pennsylvania and state courts in New Jersey, Oregon, and Tennessee. Id-, at 1104-1105, n. 9, 1109. Moreover, if we were simply to accept the Court of Appeals' conclusion, we would necessarily have to decide conclusively a federal constitutional question ( i. e., whether such an entitlement constituted property under the Due Process Clause and, if so, whether petitioner's customs or policies provided too little process to protect it). We proceed, then, to our own analysis of whether Colorado law gave respondent a right to enforcement of the restraining order.

The critical language in the restraining order came not from any part of the order itself (which was signed by the state-court trial judge and directed to the restrained party, respondent's husband), but from the preprinted notice to law-enforcement personnel that appeared on the back of the order. See supra, at 751-752. That notice effectively restated the statutory provision describing "peace officers' duties" related to the crime of violation of a restraining order. At the time of the conduct at issue in this case, that provision read as follows:

"(a) Whenever a restraining order is issued, the protected person shall be provided with a copy of such order. A peace officer shall use every reasonable means to enforce a restraining order.

"(b) A peace officer shall arrest, or, if an arrest would be impractical under the circumstances, seek a warrant for the arrest of a restrained person when the peace officer has information amounting to probable cause that:

"(I) The restrained person has violated or attempted to violate any provision of a restraining order; and

"(II) The restrained person has been properly served with a copy of the restraining order or the restrained person has received actual notice of the existence and substance of such order.

"(c) In making the probable cause determination described in paragraph (b) of this subsection (3), a peace officer shall assume that the information received from the registry is accurate. A peace officer shall enforce a valid restraining order whether or not there is a record of the restraining order in the registry." Colo. Rev. Stat. § 18-6-803.5(3) (Lexis 1999) (emphases added).

The Court of Appeals concluded that this statutory provision — especially taken in conjunction with a statement from its legislative history, and with another statute restricting criminal and civil liability for officers making arrests — established the Colorado Legislature's clear intent "to alter the fact that the police were not enforcing domestic abuse restraining orders," and thus its intent "that the recipient of a domestic abuse restraining order have an entitlement to its enforcement." 366 F. 3d, at 1108. Any other result, it said, "would render domestic abuse restraining orders utterly valueless." Id., at 1109.

This last statement is sheer hyperbole. Whether or not respondent had a right to enforce the restraining order, it rendered certain otherwise lawful conduct by her husband both criminal and in contempt of court. See §§ 18 — 6 — 803.5(2)(a), (7). The creation of grounds on which he could be arrested, criminally prosecuted, and held in contempt was hardly "valueless" — even if the prospect of those sanctions ultimately failed to prevent him from committing three murders and a suicide.

We do not believe that these provisions of Colorado law truly made enforcement of restraining orders mandatory. A well established tradition of police discretion has long co-existed with apparently mandatory arrest statutes.

"In each and every state there are long-standing statutes that, by their terms, seem to preclude nonenforcement by the police. . . . However, for a number of reasons, including their legislative history, insufficient resources, and sheer physical impossibility, it has been recognized that such statutes cannot be interpreted literally. . . . [T]hey clearly do not mean that a police officer may not lawfully decline to . . . make an arrest. As to third parties in these states, the full-enforcement statutes simply have no effect, and their significance is further diminished." 1 ABA Standards for Criminal Justice 1-4.5, commentary, pp. 1-124 to 1-125 (2d ed. 1980) (footnotes omitted).

The deep-rooted nature of law-enforcement discretion, even in the presence of seemingly mandatory legislative commands, is illustrated by Chicago v. Morales, 527 U. S. 41 (1999), which involved an ordinance that said a police officer "`shall order'" persons to disperse in certain circumstances, id., at 47, n. 2. This Court rejected out of hand the possibility that "the mandatory language of the ordinance . . . afford[ed] the police no discretion." Id., at 62, n. 32. It is, the Court proclaimed, simply "common sense that all police officers must use some discretion in deciding when and where to enforce city ordinances." Ibid, (emphasis added).

Against that backdrop, a true mandate of police action would require some stronger indication from the Colorado Legislature than "shall use every reasonable means to enforce a restraining order" (or even "shall arrest . . . or . . . seek a warrant"), §§ 18-6-803.5(3)(a), (b). That language is not perceptibly more mandatory than the Colorado statute which has long told municipal chiefs of police that they "shall pursue and arrest any person fleeing from justice in any part of the state" and that they "shall apprehend any person in the act of committing any offense . . . and, forthwith and without any warrant, bring such person before a . . . competent authority for examination and trial." Colo. Rev. Stat. § 31-4-112 (Lexis 2004). It is hard to imagine that a Colorado peace officer would not have some discretion to determine that — despite probable cause to believe a restraining order has been violated — the circumstances of the violation or the competing duties of that officer or his agency counsel decisively against enforcement in a particular instance. The practical necessity for discretion is particularly apparent in a case such as this one, where the suspected violator is not actually present and his whereabouts are unknown. Cf. Donaldson v. Seattle, 65 Wash. App. 661, 671-672, 831 P. 2d 1098, 1104 (1992) ("There is a vast difference between a mandatory duty to arrest [a violator who is on the scene] and a mandatory duty to conduct a follow up investigation [to locate an absent violator]. . . . A mandatory duty to investigate . . . would be completely open-ended as to priority, duration and intensity").

The dissent correctly points out that, in the specific context of domestic violence, mandatory-arrest statutes have been found in some States to be more mandatory than traditional mandatory-arrest statutes. Post, at 779-784 (opinion of STEVENS, J.). The Colorado statute mandating arrest for a domestic-violence offense is different from but related to the one at issue here, and it includes similar though not identical phrasing. See Colo. Rev. Stat. § 18-6-803.6(1) (Lexis 1999) ("When a peace officer determines that there is probable cause to believe that a crime or offense involving domestic violence . . . has been committed, the officer shall, without undue delay, arrest the person suspected of its commission . . ."). Even in the domestic-violence context, however, it is unclear how the mandatory-arrest paradigm applies to cases in which the offender is not present to be arrested. As the dissent explains, post, at 780-781, and n. 8, much of the impetus for mandatory-arrest statutes and policies derived from the idea that it is better for police officers to arrest the aggressor in a domestic-violence incident than to attempt to mediate the dispute or merely to ask the offender to leave the scene. Those other options are only available, of course, when the offender is present at the scene. See Hanna, No Right to Choose: Mandated Victim Participation in Domestic Violence Prosecutions, 109 Harv. L. Rev. 1849, 1860 (1996) ("[T]he clear trend in police practice is to arrest the batterer at the scene . . ." (emphasis added)).

As one of the cases cited by the dissent, post, at 783, recognized, "there will be situations when no arrest is possible, stick as when the alleged abuser is not in the home." Donaldson, 65 Wash. App., at 674, 831 P. 2d, at 1105 (emphasis added). That case held that Washington's mandatory-arrest statute required an arrest only in "cases where the offender is on the scene," and that it "d[id] not create an on-going mandatory duty to conduct an investigation" to locate the offender. Id., at 675, 831 P. 2d, at 1105. Colorado's restraining-order statute appears to contemplate a similar distinction, providing that when arrest is "impractical" — which was likely the case when the whereabouts of respondent's husband were unknown — the officers' statutory duty is to "seek a warrant" rather than "arrest." § 18-6-803.5(3)(b).

Respondent does not specify the precise means of enforcement that the Colorado restraining-order statute assertedly mandated — whether her interest lay in having police arrest her husband, having them seek a warrant for his arrest, or having them "use every reasonable means, up to and including arrest, to enforce the order's terms," Brief for Respondent 29-30. Such indeterminacy is not the hallmark of a duty that is mandatory. Nor can someone be safely deemed "entitled" to something when the identity of the alleged entitlement is vague. See Roth, 408 U. S., at 577 (considering whether "certain benefits" were "secure[d]" by rule or understandings); cf. Natale v. Ridgefield, 170 P. 3d 258, 263 (CA2 1999) ("There is no reason . . . to restrict the `uncertainty' that will preclude existence of a federally protectable property interest to the uncertainty that inheres in [the] exercise of discretion"). The dissent, after suggesting various formulations of the entitlement in question, ultimately contends that the obligations under the statute were quite precise: either make an arrest or (if that is impractical) seek an arrest warrant, post, at 785. The problem with this is that the seeking of an arrest warrant would be an entitlement to nothing but procedure — which we have held inadequate even to support standing, see Lujan v. Defenders of Wildlife, 504 U. S. 555 (1992); much less can it be the basis for a property interest. See post, at 771-772 (SOUTER, J., concurring). After the warrant is sought, it remains within the discretion of a judge whether to grant it, and after it is granted, it remains within the discretion of the police whether and when to execute it. Respondent would have been assured nothing but the seeking of a warrant. This is not the sort of "entitlement" out of which a property interest is created.

Even if the statute could be said to have made enforcement of restraining orders "mandatory" because of the domestic-violence context of the underlying statute, that would not necessarily mean that state law gave respondent an entitlement to enforcement of the mandate. Making the actions of government employees obligatory can serve various legitimate ends other than the conferral of a benefit on a specific class of people. See, e. g., Sandin v. Conner, 515 U. S. 472, 482 (1995) (finding no constitutionally protected liberty interest in prison regulations phrased in mandatory terms, in part because "[s]uch guidelines are not set forth solely to benefit the prisoner"). The serving of public rather than private ends is the normal course of the criminal law because criminal acts, "besides the injury [they do] to individuals, . . . strike at the very being of society; which cannot possibly subsist, where actions of this sort are suffered to escape with impunity." 4 W. Blackstone, Commentaries on the Laws of England 5 (1769); see also Huntington v. Attrill, 146 U. S. 657, 668 (1892). This principle underlies, for example, a Colorado district attorney's discretion to prosecute a domestic assault, even though the victim withdraws her charge. See People v. Cunefare, 102 P. 3d 302, 311-312 (Colo. 2004) (en banc) (Bender, J., concurring in part, dissenting in part, and dissenting in part to the judgment).

Respondent's alleged interest stems only from a State's statutory scheme — from a restraining order that was authorized by and tracked precisely the statute on which the Court of Appeals relied. She does not assert that she has any common-law or contractual entitlement to enforcement. If she was given a statutory entitlement, we would expect to see some indication of that in the statute itself. Although Colorado's statute spoke of "protected person[s]" such as respondent, it did so in connection with matters other than a right to enforcement. It said that a "protected person shall be provided with a copy of [a restraining] order" when it is issued, § 18-6-803.5(3)(a); that a law enforcement agency "shall make all reasonable efforts to contact the protected party upon the arrest of the restrained person," § 18-6-803.5(3)(d); and that the agency "shall give [to the protected person] a copy" of the report it submits to the court that issued the order, § 18-6-803.5(3)(e). Perhaps most importantly, the statute spoke directly to the protected person's power to "initiate contempt proceedings against the restrained person if the order [was] issued in a civil action or request the prosecuting attorney to initiate contempt proceedings if the order [was] issued in a criminal action." § 18-6-803.5(7). The protected person's express power to "initiate" civil contempt proceedings contrasts tellingly with the mere ability to "request" initiation of criminal contempt proceedings — and even more dramatically with the complete silence about any power to "request" (much less demand) that an arrest be made.

The creation of a personal entitlement to something as vague and novel as enforcement of restraining orders cannot "simply g[o] without saying." Post, at 788, n. 16 (STEVENS, J., dissenting). We conclude that Colorado has not created such an entitlement.

Even if we were to think otherwise concerning the creation of an entitlement by Colorado, it is by no means clear that an individual entitlement to enforcement of a restraining order could constitute a "property" interest for purposes of the Due Process Clause. Such a right would not, of course, resemble any traditional conception of property. Although that alone does not disqualify it from due process protection, as Roth and its progeny show, the right to have a restraining order enforced does not "have some ascertainable monetary value," as even our " Roth-type property-as-entitlement" cases have implicitly required. Merrill, The Landscape of Constitutional Property, 86 Va. L. Rev. 885, 964 (2000). Perhaps most radically, the alleged property interest here arises incidentally, not out of some new species of government benefit or service, but out of a function that government actors have always performed — to wit, arresting people who they have probable cause to believe have committed a criminal offense.

The indirect nature of a benefit was fatal to the due process claim of the nursing-home residents in O'Bannon v. Town Court Nursing Center, 447 U. S. 773 (1980). We held that, while the withdrawal of "direct benefits" (financial payments under Medicaid for certain medical services) triggered due process protections, id., at 786-787, the same was not true for the "indirect benefit[s]" conferred on Medicaid patients when the Government enforced "minimum standards of care" for nursing-home facilities, id., at 787. "[A]n indirect and incidental result of the Government's enforcement action . . . does not amount to a deprivation of any interest in life, liberty, or property." Ibid. In this case, as in O'Bannon, "[t]he simple distinction between government action that directly affects a citizen's legal rights . . . and action that is directed against a third party and affects the citizen only indirectly or incidentally, provides a sufficient answer to" respondent's reliance on cases that found government-provided services to be entitlements. Id., at 788. The O'Bannon Court expressly noted, ibid., that the distinction between direct and indirect benefits distinguished Memphis Light, Gas Water Div. v. Craft, 436 U. S. 1 (1978), one of the government-services cases on which the dissent relies, post, at 789.

We conclude, therefore, that respondent did not, for purposes of the Due Process Clause, have a property interest in police enforcement of the restraining order against her husband. It is accordingly unnecessary to address the Court of Appeals' determination ( 366 F. 3d, at 1110-1117) that the town's custom or policy prevented the police from giving her due process when they deprived her of that alleged interest. See American Mfrs. Mut. Ins. Co. v. Sullivan, 526 U. S. 40, 61 (1999).

In light of today's decision and that in DeShaney, the benefit that a third party may receive from having someone else arrested for a crime generally does not trigger protections under the Due Process Clause, neither in its procedural nor in its "substantive" manifestations. This result reflects our continuing reluctance to treat the Fourteenth Amendment as "`a font of tort law,'" Parratt v. Taylor, 451 U. S. 527, 544 (1981) (quoting Paul v. Davis, 424 U. S., at 701), but it does not mean States are powerless to provide victims with personally enforceable remedies. Although the framers of the Fourteenth Amendment and the Civil Rights Act of 1871, 17 Stat. 13 (the original source of § 1983), did not create a system by which police departments are generally held financially accountable for crimes that better policing might have prevented, the people of Colorado are free to craft such a system under state law. Cf. DeShaney, 489 U. S., at 203. The judgment of the Court of Appeals is

Reversed.

JUSTICE SOUTER, with whom JUSTICE BREYER joins, concurring.

I agree with the Court that Jessica Gonzales has shown no violation of an interest protected by the Fourteenth Amendment's Due Process Clause, and I join the Court's opinion. The Court emphasizes the traditional public focus of law enforcement as reason to doubt that these particular legal requirements to provide police services, however unconditional their form, presuppose enforceable individual rights to a certain level of police protection. Ante, at 764-765. The Court also notes that the terms of the Colorado statute involved here recognize and preserve the traditional discretion afforded law enforcement officers. Ante, at 760-764, and n. 8. Gonzales's claim of a property right thus runs up against police discretion in the face of an individual demand to enforce, and discretion to ignore an individual instruction not to enforce (because, say, of a domestic reconciliation); no one would argue that the beneficiary of a Colorado order like the one here would be authorized to control a court's contempt power or order the police to refrain from arresting. These considerations argue against inferring any guarantee of a level of protection or safety that could be understood as the object of a "legitimate claim of entitlement," Board of Regents of State Colleges v. Roth, 408 U. S. 564, 577 (1972), in the nature of property arising under Colorado law. Consequently, the classic predicate for federal due process protection of interests under state law is missing.

Gonzales implicitly recognizes this, when she makes the following argument:

"Ms. Gonzales alleges that . . . she was denied the process laid out in the statute. The police did not consider her request in a timely fashion, but instead repeatedly required her to call the station over several hours. The statute promised a process by which her restraining order would be given vitality through careful and prompt consideration of an enforcement request. . . . Denial of that process drained all of the value from her property interest in the restraining order." Brief for Respondent 10.

The argument is unconventional because the state-law benefit for which it claims federal procedural protection is itself a variety of procedural regulation, a set of rules to be followed by officers exercising the State's executive power: use all reasonable means to enforce, arrest upon demonstrable probable cause, get a warrant, and so on, see ante, at 751-752.

When her argument is understood as unconventional in this sense, a further reason appears for rejecting its call to apply Roth, a reason that would apply even if the statutory mandates to the police were absolute, leaving the police with no discretion when the beneficiary of a protective order insists upon its enforcement. The Due Process Clause extends procedural protection to guard against unfair deprivation by state officials of substantive state-law property rights or entitlements; the federal process protects the property created by state law. But Gonzales claims a property interest in a state-mandated process in and of itself. This argument is at odds with the rule that "[p]rocess is not an end in itself. Its constitutional purpose is to protect a substantive interest to which the individual has a legitimate claim of entitlement." Olim v. Wakinekona, 461 U. S. 238, 250 (1983); see also Doe v. District of Columbia, 93 F. 3d 861, 868 (CADC 1996) (per curiam); Doe v. Milwaukee County, 903 F. 2d 499, 502-503 (CA7 1990). In putting to rest the notion that the scope of an otherwise discernible property interest could be limited by related state-law procedures, this Court observed that "[t]he categories of substance and procedure are distinct. . . . `Property' cannot be defined by the procedures provided for its deprivation." Cleveland Bd. of Ed. v. Loudermill, 470 U. S. 532, 541 (1985). Just as a State cannot diminish a property right, once conferred, by attaching less than generous procedure to its deprivation, ibid., neither does a State create a property right merely by ordaining beneficial procedure unconnected to some articulable substantive guarantee. This is not to say that state rules of executive procedure may not provide significant reasons to infer an articulable property right meant to be protected; but it is to say that we have not identified property with procedure as such. State rules of executive procedure, however important, may be nothing more than rules of executive procedure.

Thus, in every instance of property recognized by this Court as calling for federal procedural protection, the property has been distinguishable from the procedural obligations imposed on state officials to protect it. Whether welfare benefits, Goldberg v. Kelly, 397 U.S. 254 (1970), attendance at public schools, Goss v. Lopez, 419 U. S. 565 (1975), utility services, Memphis Light, Gas Water Div. v. Craft, 436 U. S. 1 (1978), public employment, Perry v. Sindermann, 408 U. S. 593 (1972), professional licenses, Barry v. Barchi, 443 U. S. 55 (1979), and so on, the property interest recognized in our cases has always existed apart from state procedural protection before the Court has recognized a constitutional claim to protection by federal process. To accede to Gonzales's argument would therefore work a sea change in the scope of federal due process, for she seeks federal process as a substitute simply for state process. (And she seeks damages under Rev. Stat. § 1979, 42 U. S. C. § 1983, for denial of process to which she claimed a federal right.) There is no articulable distinction between the object of Gonzales's asserted entitlement and the process she desires in order to protect her entitlement; both amount to certain steps to be taken by the police to protect her family and herself. Gonzales's claim would thus take us beyond Roth or any other recognized theory of Fourteenth Amendment due process, by collapsing the distinction between property protected and the process that protects it, and would federalize every mandatory state-law direction to executive officers whose performance on the job can be vitally significant to individuals affected.

The procedural directions involved here are just that. They presuppose no enforceable substantive entitlement, and Roth does not raise them to federally enforceable status in the name of due process.

JUSTICE STEVENS, with whom JUSTICE GINSBURG joins, dissenting.

The issue presented to us is much narrower than is suggested by the far-ranging arguments of the parties and their amici. Neither the tragic facts of the case, nor the importance of according proper deference to law enforcement professionals, should divert our attention from that issue. That issue is whether the restraining order entered by the Colorado trial court on June 4, 1999, created a "property" interest that is protected from arbitrary deprivation by the Due Process Clause of the Fourteenth Amendment.

It is perfectly clear, on the one hand, that neither the Federal Constitution itself, nor any federal statute, granted respondent or her children any individual entitlement to police protection. See DeShaney v. Winnebago County Dept. of Social Servs., 489 U. S. 189 (1989). Nor, I assume, does any Colorado statute create any such entitlement for the ordinary citizen. On the other hand, it is equally clear that federal law imposes no impediment to the creation of such an entitlement by Colorado law. Respondent certainly could have entered into a contract with a private security firm, obligating the firm to provide protection to respondent's family; respondent's interest in such a contract would unquestionably constitute "property" within the meaning of the Due Process Clause. If a Colorado statute enacted for her benefit, or a valid order entered by a Colorado judge, created the functional equivalent of such a private contract by granting respondent an entitlement to mandatory individual protection by the local police force, that state-created right would also qualify as "property" entitled to constitutional protection.

I do not understand the majority to rule out the foregoing propositions, although it does express doubts. See ante, at 766 ("[I]t is by no means clear that an individual entitlement to enforcement of a restraining order could constitute a `property' interest"). Moreover, the majority does not contest, see ante, at 768, that if respondent did have a cognizable property interest in this case, the deprivation of that interest violated due process. As the Court notes, respondent has alleged that she presented the police with a copy of the restraining order issued by the Colorado court and requested that it be enforced. Ante, at 751, n. 1. In response, she contends, the officers effectively ignored her. If these allegations are true, a federal statute, Rev. Stat. § 1979, 42 U. S. C. § 1983, provides her with a remedy against the petitioner, even if Colorado law does not. See Cleveland Bd. of Ed. v. Loudermill, 470 U. S. 532 (1985).

The central question in this case is therefore whether, as a matter of Colorado law, respondent had a right to police assistance comparable to the right she would have possessed to any other service the government or a private firm might have undertaken to provide. See Board of Regents of State Colleges v. Roth, 408 U. S. 564, 577 (1972) ("Property interests, of course, are not created by the Constitution. Rather, they are created and their dimensions are defined by existing rules or understandings that stem from an independent source such as state law — rules or understandings that secure certain benefits and that support claims of entitlement to those benefits").

There was a time when our tradition of judicial restraint would have led this Court to defer to the judgment of more qualified tribunals in seeking the correct answer to that difficult question of Colorado law. Unfortunately, although the majority properly identifies the "central state-law question" in this case as "whether Colorado law gave respondent a right to police enforcement of the restraining order," ante, at 758, it has chosen to ignore our settled practice by providing its own answer to that question. Before identifying the flaws in the Court's ruling on the merits, I shall briefly comment on our past practice.

The majority's decision to plunge ahead with its own analysis of Colorado law imprudently departs from this Court's longstanding policy of paying "deference [to] the views of a federal court as to the law of a State within its jurisdiction." Phillips v. Washington Legal Foundation, 524 U. S. 156, 167 (1998); see also Bishop v. Wood, 426 U. S. 341, 346, and n. 10 (1976) (collecting cases). This policy is not only efficient, but it reflects "our belief that district courts and courts of appeals are better schooled in and more able to interpret the laws of their respective States." Brockett v. Spokane Arcades, Inc., 472 U. S. 491, 500-501 (1985); Hillsborough v. Cromwell, 326 U. S. 620, 629-630 (1946) (endorsing "great deference to the views of the judges of those courts `who are familiar with the intricacies and trends of local law and practice'"). Accordingly, we have declined to show deference only in rare cases in which the court of appeals' resolution of state law was "clearly wrong" or otherwise seriously deficient. See Brockett, 472 U. S., at 500, n. 9; accord, Leavitt v. Jane L., 518 U. S. 137, 145 (1996) ( per curiam).

Unfortunately, the Court does not even attempt to demonstrate that the six-judge en banc majority was "clearly wrong" in its interpretation of Colorado's domestic restraining order statute; nor could such a showing be made. For it is certainly plausible to construe " shall use every reasonable means to enforce a restraining order" and " shall arrest," Colo. Rev. Stat. §§ 18-6-803.5(3)(a)-(b) (Lexis 1999) (emphasis added), as conveying mandatory directives to the police, particularly when the same statute, at other times, tellingly employs different language that suggests police discretion, see § 18-6-803.5(6)(a) ("A peace officer is authorized to use every reasonable means to protect. . ."; "Such peace officer may transport . . ." (emphasis added)). Moreover, unlike today's decision, the Court of Appeals was attentive to the legislative history of the statute, focusing on a statement by the statute's sponsor in the Colorado House, ante, at 759, n. 6 (quoting statement), which it took to "emphasiz[e] the importance of the police's mandatory enforcement of domestic restraining orders." 366 F. 3d 1093, 1107 (CA10 2004) (en banc). Far from overlooking the traditional presumption of police discretion, then, the Court of Appeals' diligent analysis of the statute's text, purpose, and history led it to conclude that the Colorado Legislature intended precisely to abrogate that presumption in the specific context of domestic restraining orders. That conclusion is eminently reasonable and, I believe, worthy of our deference.

Even if the Court had good reason to doubt the Court of Appeals' determination of state law, it would, in my judgment, be a far wiser course to certify the question to the Colorado Supreme Court. Powerful considerations support certification in this case. First, principles of federalism and comity favor giving a State's high court the opportunity to answer important questions of state law, particularly when those questions implicate uniquely local matters such as law enforcement and might well require the weighing of policy considerations for their correct resolution. See Elkins v. Moreno, 435 U. S. 647, 662, n. 16 (1978) ( sua sponte certifying a question of state law because it is "one in which state governments have the highest interest"); cf. Arizonans for Official English v. Arizona, 520 U. S. 43, 77 (1997) ("Through certification of novel or unsettled questions of state law for authoritative answers by a State's highest court, a federal court may save `time, energy, and resources, and hel[p] build a cooperative judicial federalism'" (brackets in original)). Second, by certifying a potentially dispositive state-law issue, the Court would adhere to its wise policy of avoiding the unnecessary adjudication of difficult questions of constitutional law. See Elkins, 435 U. S., at 661-662 (citing constitutional avoidance as a factor supporting certification). Third, certification would promote both judicial economy and fairness to the parties. After all, the Colorado Supreme Court is the ultimate authority on the meaning of Colorado law, and if in later litigation it should disagree with this Court's provisional state-law holding, our efforts will have been wasted and respondent will have been deprived of the opportunity to have her claims heard under the authoritative view of Colorado law. The unique facts of this case only serve to emphasize the importance of employing a procedure that will provide the correct answer to the central question of state law. See Brockett, 472 U. S., at 510 (O'CONNOR, J., concurring) ("Speculation by a federal court about the meaning of a state statute in the absence of a prior state court adjudication is particularly gratuitous when, as is the case here, the state courts stand willing to address questions of state law on certification from a federal court").

Three flaws in the Court's rather superficial analysis of the merits highlight the unwisdom of its decision to answer the state-law question de novo. First, the Court places undue weight on the various statutes throughout the country that seemingly mandate police enforcement but are generally understood to preserve police discretion. As a result, the Court gives short shrift to the unique case of "mandatory arrest" statutes in the domestic violence context; States passed a wave of these statutes in the 1980's and 1990's with the unmistakable goal of eliminating police discretion in this area. Second, the Court's formalistic analysis fails to take seriously the fact that the Colorado statute at issue in this case was enacted for the benefit of the narrow class of persons who are beneficiaries of domestic restraining orders, and that the order at issue in this case was specifically intended to provide protection to respondent and her children. Finally, the Court is simply wrong to assert that a citizen's interest in the government's commitment to provide police enforcement in certain defined circumstances does not resemble any "traditional conception of property," ante, at 766; in fact, a citizen's property interest in such a commitment is just as concrete and worthy of protection as her interest in any other important service the government or a private firm has undertaken to provide.

In 1994, the Colorado General Assembly passed omnibus legislation targeting domestic violence. The part of the legislation at issue in this case mandates enforcement of a domestic restraining order upon probable cause of a violation, § 18-6-803.5(3), while another part directs that police officers "shall, without undue delay, arrest" a suspect upon "probable cause to believe that a crime or offense of domestic violence has been committed," § 18-6-803.6(1). In adopting this legislation, the Colorado General Assembly joined a nationwide movement of States that took aim at the crisis of police underenforcement in the domestic violence sphere by implementing "mandatory arrest" statutes. The crisis of under-enforcement had various causes, not least of which was the perception by police departments and police officers that domestic violence was a private, "family" matter and that arrest was to be used as a last resort. Sack, Battered Women and the State: The Struggle for the Future of Domestic Violence Policy, 2004 Wis. L. Rev. 1657, 1662-1663 (hereinafter Sack); id., at 1663 ("Because these cases were considered noncriminal, police assigned domestic violence calls low priority and often did not respond to them for several hours or ignored them altogether"). In response to these realities, and emboldened by a well-known 1984 experiment by the Minneapolis police department, "many states enacted mandatory arrest statutes under which a police officer must arrest an abuser when the officer has probable cause to believe that a domestic assault has occurred or that a protection order has been violated." Developments in the Law: Legal Responses to Domestic Violence, 106 Harv. L. Rev. 1498, 1537 (1993). The purpose of these statutes was precisely to "counter police resistance to arrests in domestic violence cases by removing or restricting police officer discretion; mandatory arrest policies would increase police response and reduce batterer recidivism." Sack 1670.

Thus, when Colorado passed its statute in 1994, it joined the ranks of 15 States that mandated arrest for domestic violence offenses and 19 States that mandated arrest for domestic restraining order violations. See Developments in the Law, 106 Harv. L. Rev., at 1537, n. 68 (noting statutes in 1993); N. Miller, Institute for Law and Justice, A Law Enforcement and Prosecution Perspective 7, and n. 74, 8, and n. 90 (2003), http://www.ilj.org/dv/dvvawa2000.htm (as visited June 24, 2005, and available in Clerk of Court's case file) (listing Colorado among the many States that currently have mandatory arrest statutes).

Given the specific purpose of these statutes, there can be no doubt that the Colorado Legislature used the term "shall" advisedly in its domestic restraining order statute. While "shall" is probably best read to mean "may" in other Colorado statutes that seemingly mandate enforcement, cf. Colo. Rev. Stat. § 31-4-112 (Lexis 2004) (police " shall suppress all riots, disturbances, and breaches of the peace, shall apprehend all disorderly persons in the city . . ." (emphases added)), it is clear that the elimination of police discretion was integral to Colorado and its fellow States' solution to the problem of underenforcement in domestic violence cases. Since the text of Colorado's statute perfectly captures this legislative purpose, it is hard to imagine what the Court has in mind when it insists on "some stronger indication from the Colorado Legislature." Ante, at 761.

While Colorado case law does not speak to the question, it is instructive that other state courts interpreting their analogous statutes have not only held that they eliminate the police's traditional discretion to refuse enforcement, but have also recognized that they create rights enforceable against the police under state law. For example, in Nearing v. Weaver, 295 Ore. 702, 670 P. 2d 137 (1983) (en banc), the court held that although the common law of negligence did not support a suit against the police for failing to enforce a domestic restraining order, the statute's mandatory directive formed the basis for the suit because it was "a specific duty imposed by statute for the benefit of individuals previously identified by judicial order." Id., at 707, 670 P. 2d, at 140. In Matthews v. Pickett County, 996 S. W. 2d 162 (Term. 1999) (on certification to the Sixth Circuit), the court confirmed that the statute mandated arrest for violations of domestic restraining orders, and it held that the "public duty" defense to a negligence action was unavailable to the defendant police officers because the restraining order had created a "special duty" to protect the plaintiff. Id., at 165. See also Campbell v. Campbell, 294 N. J. Super. 18, 24, 682 A. 2d 272, 274 (1996) (domestic restraining order statute "allows no discretion" with regard to arrest; "[t]he duty imposed on the police officer is ministerial"); Donaldson v. Seattle, 65 Wash. App. 661, 670, 831 P. 2d 1098, 1103 (1992) ("Generally, where an officer has legal grounds to make an arrest he has considerable discretion to do so. In regard to domestic violence, the rule is the reverse. If the officer has the legal grounds to arrest pursuant to the statute, he has a mandatory duty to make the arrest"). To what extent the Colorado Supreme Court would agree with the views of these courts is, of course, an open question, but it does seem rather brazen for the majority to assume that the Colorado Supreme Court would repudiate this consistent line of persuasive authority from other States.

Indeed, the Court fails to come to terms with the wave of domestic violence statutes that provides the crucial context for understanding Colorado's law. The Court concedes that, "in the specific context of domestic violence, mandatory-arrest statutes have been found in some States to be more mandatory than traditional mandatory-arrest statutes," ante, at 762, but that is a serious understatement. The difference is not a matter of degree, but of kind. Before this wave of statutes, the legal rule was one of discretion; as the Court shows, the "traditional," general mandatory arrest statutes have always been understood to be "mandatory" in name only, see ante, at 760. The innovation of the domestic violence statutes was to make police enforcement, not "more mandatory," but simply mandatory. If, as the Court says, the existence of a protected "entitlement" turns on whether "government officials may grant or deny it in their discretion," ante, at 756, the new mandatory statutes undeniably create an entitlement to police enforcement of restraining orders.

Perhaps recognizing this point, the Court glosses over the dispositive question — whether the police enjoyed discretion to deny enforcement — and focuses on a different question — which "precise means of enforcement," ante, at 763, were called for in this case. But that question is a red herring. The statute directs that, upon probable cause of a violation, "a peace officer shall arrest, or, if an arrest would be impractical under the circumstances, seek a warrant for the arrest of a restrained person." Colo. Rev. Stat. § 18-6-803.5(3)(b) (Lexis 1999). Regardless of whether the enforcement called for in this case was arrest or the seeking of an arrest warrant (the answer to that question probably changed over the course of the night as the respondent gave the police more information about the husband's whereabouts), the crucial point is that, under the statute, the police were required to provide enforcement; they lacked the discretion to do nothing. The Court suggests that the fact that "enforcement" may encompass different acts infects any entitlement to enforcement with "indeterminacy." Ante, at 763. But this objection is also unfounded. Our cases have never required the object of an entitlement to be some mechanistic, unitary thing. Suppose a State entitled every citizen whose income was under a certain level to receive health care at a state clinic. The provision of health care is not a unitary thing — doctors and administrators must decide what tests are called for and what procedures are required, and these decisions often involve difficult applications of judgment. But it could not credibly be said that a citizen lacks an entitlement to health care simply because the content of that entitlement is not the same in every given situation. Similarly, the enforcement of a restraining order is not some amorphous, indeterminate thing. Under the statute, if the police have probable cause that a violation has occurred, enforcement consists of either making an immediate arrest or seeking a warrant and then executing an arrest — traditional, well-defined tasks that law enforcement officers perform every day.

The Court similarly errs in speculating that the Colorado Legislature may have mandated police enforcement of restraining orders for "various legitimate ends other than the conferral of a benefit on a specific class of people," ante, at 765; see also ibid, (noting that the "serving of public rather than private ends is the normal course of the criminal law"). While the Court's concern would have some bite were we faced with a broadly drawn statute directing, for example, that the police " shall suppress all riots," there is little doubt that the statute at issue in this case conferred a benefit "on a specific class of people" — namely, recipients of domestic restraining orders. Here, respondent applied for and was granted a restraining order from a Colorado trial judge, who found a risk of "irreparable injury" and found that "physical or emotional harm" would result if the husband were not excluded from the family home. 366 F. 3d, at 1143 (appendix to dissent of O'Brien, J.). As noted earlier, the restraining order required that the husband not "molest or disturb" the peace of respondent and the daughters, and it ordered (with limited exceptions) that the husband stay at least 100 yards away from the family home. Ibid. It also directed the police to "use every reasonable means to enforce this . . . order," and to arrest or seek a warrant upon probable cause of a violation. Id., at 1144. Under the terms of the statute, when the order issued, respondent and her daughters became "`protected person[s].'" § 18-6-803.5(1.5)(a) ("`Protected person' means the person or persons identified in the restraining order as the person or persons for whose benefit the restraining order was issued"). The statute criminalized the knowing violation of the restraining order, § 18-6-803.5(1), and, as already discussed, the statute (as well as the order itself) mandated police enforcement, §§ 18-6-803.5(3)(a)-(b).

Because the statute's guarantee of police enforcement is triggered by, and operates only in reference to, a judge's granting of a restraining order in favor of an identified " `protected person,'" there is simply no room to suggest that such a person has received merely an "`incidental'" or " `indirect'" benefit, see ante, at 766-767. As one state court put it, domestic restraining order statutes "identify with precision when, to whom, and under what circumstances police protection must be afforded. The legislative purpose in requiring the police to enforce individual restraining orders clearly is to protect the named persons for whose protection the order is issued, not to protect the community at large by general law enforcement activity." Nearing, 295 Ore., at 712, 670 P. 2d, at 143. Not only does the Court's doubt about whether Colorado's statute created an entitlement in a protected person fail to take seriously the purpose and nature of restraining orders, but it fails to account for the decisions by other state courts, see supra, at 782-783, that recognize that such statutes and restraining orders create individual rights to police action.

Given that Colorado law has quite clearly eliminated the police's discretion to deny enforcement, respondent is correct that she had much more than a "unilateral expectation" that the restraining order would be enforced; rather, she had a "legitimate claim of entitlement" to enforcement. Roth, 408 U. S., at 577. Recognizing respondent's property interest in the enforcement of her restraining order is fully consistent with our precedent. This Court has "made clear that the property interests protected by procedural due process extend well beyond actual ownership of real estate, chattels, or money." Id., at 571-572. The "types of interests protected as `property are varied and, as often as not, intangible, relating `to the whole domain of social and economic fact.'" Logan v. Zimmerman Brush Co., 455 U. S. 422, 430 (1982); see also Perry v. Sindermann, 408 U.S. 593, 601 (1972) ("`[P]roperty' interests subject to procedural due process protection are not limited by a few rigid, technical forms. Rather, `property' denotes a broad range of interests that are secured by `existing rules or understandings'"). Thus, our cases have found "property" interests in a number of state-conferred benefits and services, including welfare benefits, Goldberg v. Kelly, 397 U. S. 254 (1970); disability benefits, Mathews v. Eldridge, 424 U. S. 319 (1976); public education, Goss v. Lopez, 419 U.S. 565 (1975); utility services, Memphis Light, Gas Water Div. v. Craft, 436 U. S. 1 (1978); government employment, Cleveland Bd. of Ed. v. Loudermill, 470 U. S. 532 (1985), as well as in other entitlements that defy easy categorization, see, e. g., Bell v. Burson, 402 U. S. 535 (1971) (due process requires fair procedures before a driver's license may be revoked pending the adjudication of an accident claim); Logan, 455 U. S., at 431 (due process prohibits the arbitrary denial of a person's interest in adjudicating a claim before a state commission).

Police enforcement of a restraining order is a government service that is no less concrete and no less valuable than other government services, such as education. The relative novelty of recognizing this type of property interest is explained by the relative novelty of the domestic violence statutes creating a mandatory arrest duty; before this innovation, the unfettered discretion that characterized police enforcement defeated any citizen's "legitimate claim of entitlement" to this service. Novel or not, respondent's claim finds strong support in the principles that underlie our due process jurisprudence. In this case, Colorado law guaranteed the provision of a certain service, in certain defined circumstances, to a certain class of beneficiaries, and respondent reasonably relied on that guarantee. As we observed in Roth, "[i]t is a purpose of the ancient institution of property to protect those claims upon which people rely in their daily lives, reliance that must not be arbitrarily undermined." 408 U. S., at 577. Surely, if respondent had contracted with a private security firm to provide her and her daughters with protection from her husband, it would be apparent that she possessed a property interest in such a contract. Here, Colorado undertook a comparable obligation, and respondent — with restraining order in hand — justifiably relied on that undertaking. Respondent's claim of entitlement to this promised service is no less legitimate than the other claims our cases have upheld, and no less concrete than a hypothetical agreement with a private firm. The fact that it is based on a statutory enactment and a judicial order entered for her special protection, rather than on a formal contract, does not provide a principled basis for refusing to consider it "property" worthy of constitutional protection.

Because respondent had a property interest in the enforcement of the restraining order, state officials could not deprive her of that interest without observing fair procedures. Her description of the police behavior in this case and the department's callous policy of failing to respond properly to reports of restraining order violations clearly al leges a due process violation. At the very least, due process requires that the relevant state decisionmaker listen to the claimant and then apply the relevant criteria in reaching his decision. The failure to observe these minimal procedural safeguards creates an unacceptable risk of arbitrary and "erroneous deprivation[s]," Mathews, 424 U. S., at 335. According to respondent's complaint — which we must construe liberally at this early stage in the litigation, see Swierkiewicz v. Sorema N. A, 534 U. S. 506, 514 (2002) — the process she was afforded by the police constituted nothing more than a "`sham or a pretense.'" Joint Anti-Fascist Refugee Comm. v. McGrath, 341 U.S. 123, 164 (1951) (Frankfurter, J., concurring).

Accordingly, I respectfully dissent.