Elbert Welch appeals from the judgment of the United States District Court for the Western District of New York (William M. Skretny, District Judge), dismissing his complaint alleging that officials from the New York State Department of Correctional Facilities (“DOCS”) deprived him of liberty without due process of law. The district court concluded that the deprivation in question, 90 days in solitary confinement, did not amount to an “atypical and significant hardship ... in relation to the ordinary incidents of prison life,” Sandin v. Conner, 515 U.S. 472, 484, 115 S.Ct. 2293, 132 L.Ed.2d 418 (1995), and therefore granted summary judgment for the defendants. Because we think the district court’s findings do not support this conclusion, we vacate the judgment and remand for further proceedings.
BACKGROUND
At all times relevant to this decision, Welch was an inmate at Elmira Correctional Facility, serving an indeterminate sentence of 12$ to 25 years for murder. On September 26, 1994, corrections officers discovered legal paperwork and headphones belonging to other inmates in Welch’s cell. He was charged with violating prison rules that prohibit possession of unauthorized items and provision of legal assistance without permission. Following a disciplinary hearing, Welch was found guilty of both charges and sentenced to 90 days confinement in the Special Housing Unit (“SHU”). He served most of his term in the SHU, and the balance in disciplinary keeplock in a general population cell.
Welch then brought this action under 42 U.S.C. § 1983, asserting that he had a right to provide legal assistance to prisoners and that his punishment violated due process of law for various reasons, including that he did not receive proper notice and was judged by a biased hearing officer. See Wolff v. McDonnell, 418 U.S. 539, 563-71, 94 S.Ct. 2963, 41 L.Ed.2d 935 (1974). The district court granted defendants’ motion for summary judgment, concluding that Welch did not have a right to provide legal assistance to prisoners, and that he did not have a liberty interest in being free from 90 days in the SHU because such confinement was not an “atypical and significant deprivation compared to the ordinary incidents of prison life.” Sandin, 515 U.S. at 486, 115 S.Ct. 2293. By summary order, we affirmed with respect to Welch’s asserted right to provide legal assistance, but vacated and remanded because the court had not made findings as to the nature and duration of Welch’s disciplinary confinement compared to the ordinary conditions of prison life. See Welch v. Bartlett, No. 96-2778, 1997 WL 568660 (2d Cir. Sept.12, 1997); see also Brooks v. DiFasi, 112 F.3d 46, 49 (2d Cir.1997) (requiring findings). On remand, the parties submitted affidavits, and defendants again moved for summary judgment on the ground (among others) that 90 days in the SHU was not an “atypical and significant deprivation.”
In ruling on whether Welch had suffered an atypical and significant deprivation, the district court relied on information in affidavits submitted by DOCS. Concerning conditions in the SHU, the court recognized that SHU prisoners are “confined in their cells for most of the day” and forbidden to participate in group activities. SHU prisoners can leave their cells “for one hour of exercise per day, for two showers per week, for any legal visits and one non-legal visit per week, and for occasional appointments with medical or other support staff.” By contrast, general population prisoners spend about half of each day *392locked in their cells, and are “let out for educational or vocational programming, counseling and other activities.” Regarding these activities, however, the court noted that “not every prisoner participates in such programs and it is not unusual for a prisoner’s program to be interrupted at certain times, such as when [the prisoner] transfers to a new facility.”
With respect to the duration and frequency of confinement in the SHU, the court noted that out of 215,701 inmates spending time at a DOCS facility between 1991 and 1996, a total of 19,963 were penalized with SHU confinement at least once. Thus, almost 10% of all inmates received SHU punishment. Among the prisoners confined to the SHU, 40% were confined for less than 90 days, and about 60% for 90 days or more.
Based on these facts, the court concluded that a penalty of 90 days in the SHU did not amount to an “atypical and significant hardship in relation to the ordinary incidents of prison life.” It reasoned that, while the SHU involved “a degree of deprivation,” “all DOCS prisoners suffer some degree of deprivation,” and there is “no single deprivation that is unique to SHU.” Furthermore, “a significant proportion of DOCS inmates spend at least some time in SHU,” and Welch’s term was not unusual. The court therefore granted summary judgment for defendants. Welch appealed.
DISCUSSION
In Hewitt v. Helms, 459 U.S. 460, 103 S.Ct. 864, 74 L.Ed.2d 675 (1983), and related cases, the Supreme Court ruled that when a prisoner brings an action under § 1983 asserting a due process right that is premised on a state-created liberty interest, the prisoner must establish that the state’s laws in fact create such a liberty interest.1 See Sandin, 515 U.S. at 479-81, 115 S.Ct. 2293. The Court instructed that such a liberty interest arises when state statutes or regulations require, in “language of an unmistakably mandatory character,” that a prisoner not suffer a particular deprivation absent specified predicates. See Hewitt, 459 U.S. at 471-72, 103 S.Ct. 864.
In Sandin, however, the Court held that a mandatory obligation of prison officials for the prisoner’s benefit is insufficient in itself to create a due process right enforceable by an action under § 1983. A liberty interest must also be such that its deprivation would subject the prisoner to “atypical and significant hardship ... in relation to the ordinary, incidents of prison life.” Sandin, 515 U.S. at 484, 115 S.Ct. 2293. Thus, Sandin held that thirty days in punitive solitary confinement did not deprive a Hawaii prisoner of a protected liberty interest, where that term “mirrored those conditions imposed upon inmates in administrative segregation and protective custody,” the conditions were “within the [expected] contour of the actual sentence imposed,” id. at 486 & n. 9, 115 S.Ct. 2293, and the challenged confinement was not significantly onerous compared to the “ ‘lockdown time’ even for inmates in the general population.” Id. & n. 8.
After Sandin, a prisoner who experiences a deprivation arising under mandatory rules has no actionable due process claim if other prisoners experience approximately the same deprivation in the ordinary administration of the prison with sufficient regularity that such deprivation is typical. See, e.g., Frazier v. Coughlin, 81 F.3d 313, 317-18 (2d Cir.1996). In other words, actions under the Due Process Clause are reserved for prisoners enduring a hardship that is substantially more grave than hardships they would be likely to endure simply as a consequence of the ordinary administration of the prison. *393Where a statute limits a prison’s ability to impose a constraint as a punishment, but the prison makes a practice of imposing for non-punitive reasons a constraint endured under similar conditions and for a similar duration with sufficient regularity, then freedom from the deprivation is not a right of “real substance” which due process protects. See Sandin, 515 U.S. at 478, 480, 115 S.Ct. 2293.
The district court reasonably separated the issue of the severity of the deprivation imposed on the plaintiff from its duration. See Arce v. Walker, 139 F.3d 329, 337 (2d Cir.1998). We believe, however, that the court’s analysis was flawed in certain respects.
First, the court apparently assumed that the conditions of confinement for SHU prisoners were not qualitatively different from the conditions endured by general population prisoners when confined to their cells. This assumption was not justified by the record. Welch’s affidavit states that conditions of hygiene are far inferior for SHU prisoners than for the general population. He alleges that SHU prisoners, unlike general population, receive inadequate amounts of toilet paper, soap and cleaning materials, a filthy mattress, and infrequent changes of clothes. The affidavit submitted by DOCS did not discuss comparability of these conditions. The court was obligated to accept Welch’s allegations for purposes of the motion for summary judgment and could not properly assume the conditions to be equivalent. See Burgio & Campofelice, Inc. v. NYS Dep’t of Labor, 107 F.3d 1000, 1005 (2d Cir.1997). The question of course remains whether the differences are of such nature as to constitute a “significant hardship,” Sandin, 515 U.S. at 484, 115 S.Ct. 2293, that is atypical.
Second, while recognizing that SHU prisoners are confined to their cells for close to 23 hours a day, cannot participate in group activities, and have less access than is normal for general population prisoners to showers, visits, and other privileges, the court discounted these differences for two reasons. First, prisoners in general population “typically spend about half of every day locked in their cells;” second, with respect to out-of-cell programs, “not every prisoner participates in such programs and it is not unusual for a prisoner’s program to be interrupted at certain times, such as when [the prisoner] transfers to a new facility.” Welch v. Bartlett, No. 94-CV-718S, slip op. at 12-13 (W.D.N.Y. July 20, 1998).
In our view, these facts do not justify the court’s conclusion that the conditions of SHU confinement (considered without regard to duration or frequency of imposition) are not “atypical” compared to the conditions of general population confinement. Although confinement to one’s cell for half the day has some similarity to such confinement for 23 hours a day, the difference seems to us to be great. Furthermore, the fact that general population prisoners’ access to -programs is sometimes restricted or interrupted does not show either that such limitations occur with sufficient regularity to be considered typical, or that the severity of the conditions faced by a prisoner experiencing such limitations on his programs is comparable. Whether the conditions of Welch’s confinement constitute an atypical and significant hardship requires that they be considered in comparison to the hardships endured by prisoners in general population, as well as prisoners in administrative and protective confinement, assuming such confinements are imposed in the ordinary course of prison administration. Further, the duration and the frequency of such deprivations are highly relevant to whether the conditions of a plaintiffs confinement should be considered atypical. The court did not consider the frequency or duration of non-punitive confinements such as administrative and punitive segregation, keeplock and cube confinements. Cf. generally Lee v. Coughlin, 26 F.Supp.2d 615, 618-629 (S.D.N.Y.1998) (setting out comparative *394statistics and describing conditions in various forms of confinements).
Finally, we believe the court erred in concluding that the 90-day duration of Welch’s confinement in the SHU did not render it atypical because approximately half the punitive SHU sentences were 90 days or more. In our view, the relevant comparison concerning duration is between the period of deprivation endured by the plaintiff and periods of comparable deprivation typically endured by other prisoners in the ordinary course of prison administration, including general population prisoners and those in various forms of administrative and protective custody. The theory of Sandin is that, notwithstanding a mandatory entitlement, a deprivation is not of sufficient gravity to support a claim of violation of the Due Process Clause if similar deprivations are typically endured by other prisoners, not as a penalty for misbehavior, but simply as the result of ordinary prison administration. See Sandin, 515 U.S. at 486, 115 S.Ct. 2293. The comparison required by Sandin therefore is not between the duration of plaintiffs SHU sentence and the SHU terms received by others who were convicted of misbehavior. That comparison does not tell whether Welch’s deprivation was more serious than typically endured by prisoners as an ordinary incident of prison life.
For the same reasons, we are troubled by the court’s reliance on the fact that 10% of the prison population receives punitive terms in the SHU for misbehavior at some point during the service of their sentences. Punitive terms in the SHU, though doubtless warranted for some forms of misbehavior, are not a “normal incident” for a prisoner whose wrongdoing must be established according to due process standards if the consequence of an adverse finding is confinement in atypical conditions of severe hardship. How many prisoners receive such terms as punishment for misbehavior does not measure how likely a prisoner is to suffer comparable privation in the ordinary administration of the prison.2
For the foregoing reasons, we conclude that the district court erred in finding that Welch failed to raise a genuine question of material fact whether his punishment constituted an atypical and significant deprivation under Sandin. Nor, however, can we say on this record that his punishment did constitute such a deprivation. The record does not reveal whether it is typical for inmates not being disciplined to spend similar periods of time in similar circumstances.3 See Brooks, 112 F.3d at 49.
We vacate the summary judgment for defendants and remand for further proceedings.4 On remand, unless the defen*395dants prevail on summary judgment with respect to issues other than the existence of a liberty interest, there must either be a trial of the Sandin issue or the presentation by the defendants of sufficient undisputed facts to demonstrate that summary judgment is warranted in the light of this opinion.
CONCLUSION
The judgment of the district court is vacated, and the case is remanded for further proceedings consistent with this opinion.5
. A different case is presented when a prisoner asserts that a liberty interest arises from "the Due Process Clause of its own force.” Sandin, 515 U.S. at 484, 115 S.Ct. 2293 (citing Vitek v. Jones, 445 U.S. 480, 493, 100 S.Ct. 1254, 63 L.Ed.2d 552 (1980)).
. Even if we concluded that this statistic were pertinent, we wonder whether the fact that 10% of the prison population received a SHU confinement (while 90% did not), and 6% received a term of SHU confinement in excess of 90 days (while 94% did not) would make such a confinement typical of the ordinary incidents of prison life.
. We do not imply by this opinion that summary judgment in favor of DOCS might not ultimately be found appropriate, either under Sandin or on other grounds. See Sandin, 515 U.S. at 491, 115 S.Ct. 2293 (Ginsburg, J., dissenting) (noting that, had the Court denied summary judgment because the prisoner in fact was deprived of a liberty interest, a motion for summary judgment would have been in order on the ground that the prisoner was afforded all the process he was due).
. We reject the defendants’ further contention that under Sandin, SHU confinement cannot give rise to a protected liberty interest. Without clearer indication from the Supreme Court, Hewitt, as modified by Sandin 's additional requirement of an "atypical and significant hardship,” remains good law in analyzing whether statutory or regulatory tests can generate a liberty interest. See Frazier, 81 F.3d at 317-18. Because the regulations governing punitive confinement in the SHU are the same as those we found sufficient to create a liberty interest prior to Sandin, see Walker v. Bates, 23 F.3d 652, 655-656 (2d Cir.1994), we conclude that defendants were not entitled to summary judgment on the ground that state law did not create a liberty interest in not being confined to the SHU.
. If the district court proceeds to rule on the Sandin issue, the court should consider procuring counsel for the plaintiff. In an early case that may establish binding law as to whether periods in the SHU deprive prisoners of liberty under the Due Process Clause, it is important for the court to be presented with a full and accurate picture of the comparison between SHU confinement and the ordinary incidents of prison life. An unrepresented prisoner may be incapable of presenting all the pertinent evidence on his side.