dissenting.

I

Only the majority’s unjustifiably broad reading of the Arizona Supreme Court’s decision in Mileham serves as an arguable basis for its validation of the state’s constitutionally impermissible attempt to restrict Appellant’s eligibility for parole. A more careful and restrained analysis of Mileham leads to the inescapable conclusion that the State’s conduct in this case violates the Constitution’s prohibition against ex post facto punishment.

The ex post facto clause of the Constitution forbids legislation that operates to the “substantial disadvantage” of prisoners, whether or not the legislation is “technically an increase in the punishment annexed to the crime.” Lindsey v. Washington, 301 U.S. 397, 401-02 , 57 S.Ct. 797, 799 , 81 L.Ed. 1182 (1937). In Love v. Fitzharris, 460 F.2d 382 (9th Cir.1972), vacated as moot, 409 U.S. 1100 , 93 S.Ct. 896 , 34 L.Ed.2d 682 (1973), we emphasized that application of the ex post facto clause is not limited to acts of legislation. On the contrary, “[a] new administrative interpretation which subjects the prisoner already sentenced to more severe punishment has the same effect as a new statute lengthening his present term or a new court decision making what was lawful when done a) crime .... [E]ach is prohibited by the Constitution.” 460 F.2d at 385 *376 (citations omitted). The interpretation of the statute by the administrative agency charged with its enforcement has “the force and effect of law.” Id. Defendants are entitled to rely on administrative interpretations, including parole eligibility guidelines, when deciding whether to enter a guilty plea, 1 and trial judges consider such guidelines in concluding the sentence to be imposed. See Weaver v. Graham, 450 U.S. 24, 32 , 101 S.Ct. 960, 966 , 67 L.Ed.2d 17 (1981).

Our decision in Love v. Fitzharris requires that appellant’s habeas corpus petition be granted. The facts of Love are identical in all important respects to those presented here. In Love, the administrative agency responsible for parole guidelines changed its method of computing parole eligibility after the petitioner was sentenced. The changes thus altered the legal consequences of acts committed before the new parole guidelines became effective. Consequently, we held that the application of the new parole regulations to the petitioner violated the constitution’s prohibition against ex post facto punishment. The result should be the same in the case before us.

In the past we have said that there is a limited exception to the Love rule, and that increased punishment resulting from a new court pronouncement on a statute already in existence when a defendant was sentenced might not, under certain circumstances, constitute ex post facto punishment. See Mileham v. Simmons, 588 F.2d 1279, 1280 (9th Cir.1979). 2 The majority seizes upon this limited, and questionable, exception and holds that the new parole eligibility guidelines were the direct result of the Arizona Supreme Court’s pronouncement in Mileham on an already existing statute. In doing so, the majority expands Mileham far beyond its holding and its intended scope.

In my view, the majority significantly rewrites the Arizona Supreme Court’s decision and applies it to new issues and circumstances not before the Arizona court and to a different statute than the one construed by the court in that case. It does so after acknowledging that appellant must prevail if we are required to “extend” Mileham in order to rule against him. Yet Mileham decided only a single, very limited question: How should parole eligibility be determined for a prisoner convicted of an escape attempt under A.R.S. § 13-392? The court’s explicitly circumscribed holding is that “a sentence for escape runs consecutive to an original sentence and therefore can commence only when the first sentence has been completely served.” Mileham v. Arizona Board of Pardons and Paroles, 110 Ariz. 470, 471 , 520 P.2d 840, 841 (1974). Nevertheless, the majority interprets Mile-ham as altering the way in which all consecutive sentencing statutes should be interpreted while claiming that it is merely applying, not extending, the Mileham holding. For four reasons, the majority’s attempt to characterize its extension of Mileham as simply an application of a prior holding is unjustified.

First, the Arizona Supreme Court clearly limits its holding to the escape statute. Although some of the language in Mileham suggests that other consecutive sentencing statutes might be interpreted in the same way, the holding is not expanded beyond *377 the escape statute. Indeed, the court relies exclusively on the escape statute in justifying its decision. The court emphasizes that “the Legislature intended to increase the punishment of those who attempted to or escaped from the State Prison.” Id. at 472, 520 P.2d at 842 . As a result, the parole guidelines for prisoners convicted of escape had to be altered: “We conclude that the Legislature, by providing in the escape statute . .. that a prisoner’s term of imprisonment for escape commence at the time he would otherwise have been discharged from the prison, foreclosed any possibility that the prisoner may be released on parole.” Id. at 473 , 520 P.2d at 843 . Moreover, the only precedent cited by the court to support its holding, Ex parte Irwin, 88 Cal. 169 , 25 P. 1118 (1891), involves a similar California escape statute. In short, there should be no doubt that the Mileham holding is carefully limited to parole eligibility for prisoners sentenced under the escape statute.

Second, this court has already recognized the limited nature of the Arizona Supreme Court’s holding. In denying Mileham’s habeas corpus petition, we held that the escape statute, section 13-392, was “in effect when Mileham was convicted” of escape and that the Arizona Supreme Court had “now authoritatively construed the statute.” Mileham v. Simmons, 588 F.2d 1279, 1280 (9th Cir.1979). In 1979, then, we did not find that the Arizona Supreme Court’s Mileham holding applied to anything more than the escape statute, and there is no justification for our reaching a contrary result today.

Third, the Arizona Supreme Court was surely aware of the way parole eligibility guidelines were being interpreted at the time of its Mileham decision. If the court believed these interpretations were incorrect, it could easily have said so in Mileham. Instead of correcting the existing interpretations, however, the Arizona Supreme Court unequivocally limited its holding to the escape statute.

Fourth, the Arizona Supreme Court’s decision in Mileham did not have the immediate effect of altering the administrative interpretation. On the contrary, the Arizona Attorney General did not change the parole eligibility guidelines until several years after the Mileham decision. 3 In the meantime, the state continued to apply the old guidelines. This significant delay is convincing proof that the Mileham holding was not perceived by Arizona officials, at least for a substantial period of time, as affecting anything more than the escape statute. Moreover, it demonstrates the fact that it was a subsequent change in administrative interpretation, rather than the Mile-ham decision alone, that caused the increased punishment in appellant’s case.

In sum, the majority’s reading of Mile-ham is completely unwarranted; the majority is extending not applying that decision. Because the Arizona Supreme Court’s holding was limited only to the escape statute, the subsequent adverse administrative alterations in the parole guidelines cannot be constitutionally applied to prisoners sentenced before the changes were adopted. Appellant’s imprisonment under parole guidelines that increase the time he must spend in prison before he is eligible for release and that were not in effect at the time he was sentenced offends the Constitution’s prohibition against ex post facto punishment.

II

There is a second and independent reason why I believe the majority’s decision is incorrect. I have assumed in the preceding part of my dissent that the Arizona Supreme Court’s holding in Mileham and our subsequent decision denying Mileham’s habeas corpus petition are still valid despite the Supreme Court’s recent decision in Weaver v. Graham, 450 U.S. 24 , 101 S.Ct. 960 , 67 L.Ed.2d 17 (1981). This assumption, which the majority accepts without question, is not necessarily warranted.

*378 In Weaver, the Supreme Court considered a Florida law that reduced the amount of “gain time” for good conduct that could be deducted from a prisoner’s sentence. The Court held that the statute was unconstitutional as an ex post facto law when applied to a prisoner whose crime was committed before the statute was enacted. 450 U.S. at 28-36 , 101 S.Ct. at 964-968 . The Court’s holding calls the two Mileham decisions into serious question.

To begin with, the argument that Mile-ham had no “vested right” in an erroneous interpretation of an existing state statute is essential to the two Mileham decisions. See, e.g., Mileham v. Simmons, 588 F.2d 1279, 1280 (9th Cir.1979). In Weaver, however, the Supreme Court held that such vested rights considerations are not relevant: “When a court engages in ex post facto analysis, which is concerned solely with whether a statute assigns more disadvantageous criminal or penal consequences to an act than did the law in place when the act occurred, it is irrelevant whether the statutory change touches any vested rights.” 450 U.S. at 30 n. 13, 101 S.Ct. at 964 n. 13. As the Seventh Circuit Court of Appeals recently affirmed, “the prisoner does not have to show that he had a vested right to be paroled. That showing would be necessary for a contract clause or due process challenge, but it is not relevant to an ex post facto claim.” Welsh v. Mizell, 668 F.2d 328, 332 (7th Cir.1982).

Next, the Weaver decision emphasizes that what is “[cjritical to relief” under an ex post facto analysis is not “an individual’s right to less punishment” but “the lack of fair notice.” 450 U.S. at 30 , 101 S.Ct. at 965 . There is no dispute that the right to fair notice is violated when the legislature alters the punishment after a defendant has committed a criminal act. Nor is there any dispute that fair notice is denied when there is an increase in a defendant’s punishment as the result of a subsequent administrative interpretation of parole guidelines. It is the same right to fair notice that is violated when a court subsequently reinterprets a statute or invalidates an administrative interpretation under which the defendant was sentenced and, as a result, his punishment is made more severe. 4 The fact that a court, rather than a legislature, is changing the interpretation of an existing statute, or a court, rather than an agency, is invalidating a prior administrative practice, appears to be wholly irrelevant under the principles enunciated in Weaver. In all of these instances, the new interpretations are “both retrospective and more onerous than the law in effect on the date of the offense.” 450 U.S. at 30-31 , 101 S.Ct. at 965 (footnote omitted). At least when the defendant, in deciding how to plead, and the judge, in calculating the sentence to be imposed, both rely on existing law, it seems to me no longer possible in light of Weaver to draw constitutionally significant distinctions between post facto judicial pronouncements and legislative enactments.

The principal argument for distinguishing subsequent court interpretations from new statutes or administrative interpretations is that the former, or at least some of the former, may be “foreseeable” and the latter are not. See Mileham, 588 F.2d at 1280 . This always somewhat dubious distinction, however, is wholly unpersuasive after Weaver . The decisive question, according to the Supreme Court, is whether the altered statute or interpretation “changes the legal consequences of acts completed before its effective date.” 450 U.S. at 31 , 101 S.Ct. at 965 . 5 There is no doubt that the effect of the new statutory interpretation here is to increase the punishment beyond that imposed by the trial judge during sentencing. In this context it makes little sense to argue that defendants and sentencing judges could or should “foresee” possible judicial changes in statutory interpretation. On the contrary, the existing law “is a significant factor enter *379 ing into both the defendant’s decision to plea bargain and the judge’s calculation of the sentence to be imposed.” 450 U.S. at 32 , 101 S.Ct. at 966 . A defendant is entitled to rely on that existing law when deciding whether to plead guilty, just as is the sentencing judge when deciding what sentence he will impose. To say that they may so rely — except to the extent that the courts may subsequently change the rules— would create a substantial conflict with the Weaver principles.

Under Weaver , an existing administrative interpretation may not be changed retroactively to the substantial disadvantage of a prisoner. In my view, any such alteration in that interpretation — even if it results from a court pronouncement — may not constitutionally be applied to those already sentenced. Thus, Weaver provides an alternative reason why Appellant’s habeas corpus petition should be granted.

I would grant the relief requested.

. Appellant Holquin pleaded guilty to two crimes in January 1973.

. The ex post facto clause limits the powers of legislatures but does not, of its own force, apply to the judicial branch. See, e.g., Marks v. United States, 430 U.S. 188, 191 , 97 S.Ct. 990, 992 , 51 L.Ed.2d 260 (1977). Nonetheless, the principle of fair warning that is the foundation of the ex post facto clause does restrict the retroactive application of judicial decisions. Marks, 430 U.S. at 191, 97 S.Ct. at 992 ; Forman v. Wolff, 590 F.2d 283, 284 (9th Cir.1978).

For that reason, we have analyzed judicial decisions under the ex post facto clause in similar cases. See, e.g., Mileham, 588 F.2d at 1280 ; Forman, 590 F.2d at 284 .

Because of the United States Supreme Court’s recent application of the fair warning principle in Weaver v. Graham, 450 U.S. 24 , 101 S.Ct. 960 , 67 L.Ed.2d 17 (1981), I am doubtful that the limited exception we previously created in our Mileham opinion is still valid. See infra p. 374 .

. Although the Arizona Supreme Court’s Mile-ham decision was handed down in 1974, the Attorney General did not alter the parole guidelines until November 15, 1977.

. See supra note 2.

. We note that under Weaver the relevant date for determining whether an unconstitutional increase in punishment has occurred is the date of the offense, not the date of sentencing. 450 U.S. at 33 , 101 S.Ct. at 966 .