concurring and dissenting:
I concur in the majority opinion, with the exception of part III-A and the second sentence of the second paragraph of part III-B. As to those three portions, I dissent.
A. Burdens of Production and Persuasion
Andrews asserts that the district court erred when it allocated to him the burden of proving that his prior dismissals did not constitute strikes. He posits that when the state tells the court that a prisoner should not have IFP status based on 28 U.S.C. § 1915(g), it should bear the final burden of persuading the court of the existence of three prior strikes. The majority and I agree that is not so.
The district court’s allocation of the burden of persuasion is a question of law that we review de novo. Ferrari, Alvarez, Olsen & Ottoboni v. Home Ins. Co., 940 F.2d 550, 555 (9th Cir.1991); Lew v. Moss, 797 F.2d 747, 750 (9th Cir.1986). Although we have yet to rule on the issue, those circuits that have addressed it have placed the burden on the prisoner who is seeking to file IFP. See Day v. Maynard, 200 F.3d 665, 666-67 (10th Cir.1999) (per curiam) (section 1915(g) barred prisoner’s action where he failed to persuade the court that his prior dismissals did not qualify as strikes); Evans v. Ill. Dep’t of Corrs., 150 F.3d 810, 812 (7th Cir.1998) (“Having been notified [of the court’s denial of his request to proceed IFP], Evans bore the burden of showing that the district court incorrectly assessed his litigation history.”); Rivera v. Allin, 144 F.3d 719, 730 (11th Cir.1998) (“[I]t is the prisoner’s burden to produce sufficient record information about ‘prior occasions’ to the appellate court....”). I join them. So, as I understand it, does the majority.
Where the majority and I disagree is on just what a defendant or the district court must point to in order to require the plaintiff to go forward and shoulder his burdens of production and persuasion.
In general, filing an action IFP is a privilege, not a right. See, e.g., Rodriguez v. Cook, 169 F.3d 1176, 1180 (9th Cir.1999); *1124White v. Colorado, 157 F.3d 1226, 1233 (10th Cir.1998); Rivera, 144 F.3d at 724; Startti v. United States, 415 F.2d 1115, 1116 (5th Cir.1969) (per curiam); Smart v. Heinze, 347 F.2d 114, 116 (9th Cir.1965). Therefore, prisoners who desire that benefit have always been statutorily required to demonstrate that they financially qualify,1 and with § 1915(g) Congress added one more hurdle that they must surmount. To place the burden of proving the existence of (or establishing a prima facie case of) three strikes on the state would be illogical for it would force a third party to demonstrate why a person should not obtain a benefit to which the person has no right in the first place. See, e.g., Aeroquip Corp. v. Aetna Cas. & Sur. Co., Inc., 26 F.3d 893, 895 (9th Cir.1994) (allocating burden of persuasion to insured in part because it aligned the burden with the benefit).2 The more sound approach requires the party seeking IFP status to prove all of the prerequisites thereto, including the absence of three prior strikes. Under that framework, a prisoner has an incentive to file a motion for IFP status only if he knows that he can produce sufficient evidence to show that his prior dismissals do not qualify under § 1915(g), thus reducing frivolous IFP motions and effectuating the policy underlying the PLRA. See Howard, 894 F.2d at 1090.
But what about access to the relevant records? Andrews opines that the state is better situated to ascertain and provide courts with information regarding prisoners’ prior dismissals. To bolster his assertion, he relies heavily on our pronouncement that, in the administrative exhaustion context, “prison officials are likely to have greater legal expertise and, as important, superior access to prison administrative records in comparison to prisoners,” particularly where prisoners, have been transferred to different facilities. Wyatt v. Terhune, 315 F.3d 1108, 1119 (9th Cir.2003).
Contrary to Andrews’ intimation, although easier access to records can be relevant to the burden-allocation inquiry, it is not dispositive.3 See, e.g., NLRB v. Tahoe Nugget, Inc., 584 F.2d 293, 301 (9th Cir.1978). It was, in fact, only one amongst several factors underlying Wyatt. More importantly, evidence of prior dismissals is easily distinguishable from evidence of prison administrative records. The latter are likely produced by and stored at a prison facility; therefore, prison administrators are intimately familiar with them, and they are easily identifiable and accessible. As evidenced by this case, however, prisoners may file legal actions in numerous courts and under different names. Copies of those dismissals and the details of the nature of those cases would at least presumptively be more readily ascertained by the prisoner, who filed them and followed them to their conclusion, than they are by a third party’s attorney. Cer*1125tainly, one who files a lawsuit can be expected to know more about it than some third party.4 Moreover, the fact that the state brought the prior actions to the attention of the court should make no difference at all. The question of IFP status is a matter between the court and the re-questor. The mere fact that the state alerts the court to the issue (and might benefit from its resolution) should not lead to the placing of any substantial burden upon it.5
In sum, the goals and access factors as well as the policies underlying § 1915(g) militate in favor of placing the burden of persuasion on Andrews. More than that, it should be enough for the defendants or the court to point to the fact of prior dismissals and then let the plaintiff explain them, if he can. That is, the dismissals themselves are sufficient evidence to suggest that, at the least, the action was not meritorious and failed to state a claim. The plaintiff should have to demonstrate the contrary. Thus, I would hold that when the state proffers information tending to show that a prisoner has three prior strikes, it is the prisoner’s burden to produce sufficient evidence to persuade the court that § 1915(g) does not bar him from IFP status. Moreover, the dismissals themselves tend to show just that.
B. The Meaning of a Strike
I also do not agree that a dismissal for filing a clearly improper appeal is not a dismissal on the basis of frivolity. This court, for example, is often barraged with premature appeals by prisoners who refuse to accept the district court’s interim rulings- — for example, dismissals with leave to amend, which we have definitively stated are not appealable. See WMX Techs., Inc. v. Miller, 104 F.3d 1133, 1136-37 (9th Cir.1997) (en banc); see also Neitzke v. Williams, 490 U.S. 319, 325, 109 S.Ct. 1827, 1831-32, 104 L.Ed.2d 338 (1989); Tripati v. First Nati Bank & Trust, 821 F.2d 1368, 1370 (9th Cir.1987). Indeed, as I see it, that sort of a filing is close to the apogee of filings which meet the definition set forth in the majority opinion.
Thus, I concur in the majority opinion, except as to the portions already indicated, as to which I respectfully dissent.
. 28U.S.C. § 1915(a)(1).
. Allocating the burden of persuasion to the benefit-seeker has been common in other areas as well. For example, in the sentencing context, the burden of proving facts that would warrant a reduced sentence rests on the defendant. See Butler, 970 F.2d at 1026; United States v. Howard, 894 F.2d 1085, 1089-90 (9th Cir.1990); see also United States v. Garcia, 544 F.2d 681, 685-86 (3d Cir.1976) ("One who affirmatively seeks special favor at sentencing has the burden of proving why it should be bestowed."). That is also true for disability-benefit claimants see Gomez v. Chater, 74 F.3d 967, 970 (9th Cir.1996), as well as for tax-deduction suppliants, see Norgaard v. Comm’r, 939 F.2d 874, 877 (9th Cir.1991).
.Actually, leading commentators have cautioned against its overemphasis. See McCormick, supra, § 337, at 413; Richard A. Epstein, Pleadings and Presumptions, 40 U. Chi. L.Rev. 556, 579-80 (1973); Edward W. Cleary, Presuming and Pleading: An Essay on Juristic Immaturity, 12 Stan. L.Rev. 5, 12 (1959).
. By the way, Andrews should not be able to hide behind the claim that because his litigation history is such a gallimaufry (over 60 cases in all), he should not have to remember or explain that history. That is the height of crocodility; he is just the sort of prisoner that Congress had in mind when it passed the PLRA.
. In fact, it is difficult to see why the state should even have any burden of production under the circumstances. It aids and is a friend of the court when it brings the information to the court's attention. After all, filing fees are designed to fund court operations; they do not go to the opposing litigants. See 28 U.S.C. §§ 1911-1931.