Mitchell Marbury v. Warden III
CourtCourt of Appeals for the Eleventh Circuit
Date FiledJuly 7, 2026
Docket24-13513
StatusPublished
📰 News Coverage: Read the LAWS.com news report on this case
Full Opinion
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 1 of 22
FOR PUBLICATION
In the
United States Court of Appeals
For the Eleventh Circuit
____________________
No. 24-13513
____________________
MITCHELL MARBURY,
Plaintiff-Appellant,
versus
WARDEN III,
WARDEN II,
CAPT. CARLA GRAHAM,
Defendants-Appellees,
BEVERLY WARREN
C.O.,
Defendant.
____________________
Appeal from the United States District Court
for the Northern District of Alabama
D.C. Docket No. 4:18-cv-00925-CLS-JHE
____________________
Before NEWSOM, KIDD, and WILSON, Circuit Judges.
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 2 of 22
2 Opinion of the Court 24-13513
NEWSOM, Circuit Judge:
In the last few years, St. Clair Correctional Facility in Ala-
bama has been the subject of a number of lawsuits in which in-
mates have alleged that the threat of prisoner-on-prisoner violence
is so grave that it risks violating the Eighth Amendment. See, e.g.,
Boykins v. Dunn, 696 F. Supp. 3d 1061, 1066, 1072 (N.D. Ala. 2023)
(denying prison officials summary judgment where the record re-
flected that the facility suffered from “chronic understaffing, non-
compliance with institutional policies, inadequate locks, cameras,
and metal detectors, and minimal control over inmate move-
ment,” and prisoners readily accessed weapons by “fashion[ing]
knives from the chain link fence, bed rails, Coke cans, medical
splints, or any other piece of metal [they could] access”); McCarley
v. Dunn, 722 F. Supp. 3d 1242, 1251–52 (N.D. Ala. 2024) (denying
motion to dismiss where an inmate asserted that violence at the
prison “far exceeded typical levels” at comparable facilities and
cited “insufficient staffing and supervision; unauthorized inmate
movement throughout the prison; and the proliferation of contra-
band weapons” (internal citations omitted)); Wilson v. Dunn, 618 F.
Supp. 3d 1253, 1264 (N.D. Ala. 2022) (denying motion to dismiss
where a prisoner alleged “widespread contraband weapons, under-
staffing, poor inmate supervision, and unrestricted prisoner move-
ment among cell blocks . . . result[ing] in hundreds of violent inci-
dents at St. Clair per year”). This case is another in that line.
St. Clair inmate Mitchell Marbury appeals the district court’s
grant of summary judgment in favor of Warden DeWayne Estes,
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 3 of 22
24-13513 Opinion of the Court 3
Warden Cedric Specks, and Captain Carla Graham on his claim
that they were deliberately indifferent to his safety in violation of
the Eighth Amendment. The district court held that Marbury had
failed to show a genuine dispute of material fact as to whether he
faced a “substantial risk of serious harm” in the prison. After care-
ful consideration, and with the benefit of oral argument, we vacate
the district court’s decision and remand for further proceedings.
I
We’ll summarize the pertinent background in three parts.
First, we’ll describe an earlier deliberate-indifference case brought
by Marbury, Marbury v. Warden, 936 F.3d 1227 (11th Cir. 2019) (per
curiam) (Marbury I), which guides our analysis here. Next, we’ll
recount the specific factual allegations underlying this case. And
finally, we’ll describe the winding procedural history that led to this
appeal.
A
Our decision in Marbury I reiterated and formalized guide-
lines for analyzing an Eighth Amendment deliberate-indifference
claim that is based not on a specific and identifiable threat but, ra-
ther, on the theory that a “generalized risk of violence” pervades a
prison. 936 F.3d at 1235. That case, like this one, centered on Mar-
bury’s experience at St. Clair. Id. at 1231. In 2016, Marbury asked
prison officials to transfer him from one dorm to another because
he felt unsafe, but, by his account, they refused—and even
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 4 of 22
4 Opinion of the Court 24-13513
mocked—his requests. 1 Id. at 1231–32. Not long thereafter, Mar-
bury was attacked and stabbed by another inmate. Id. at 1232.
Proceeding pro se, Marbury brought a 42 U.S.C. § 1983
claim alleging that the prison’s officials were “deliberately indiffer-
ent to a substantial risk to his safety.” Id. at 1231. The district court
granted summary judgment to the officials. This Court affirmed
on the ground that Marbury’s evidence “regarding a general risk of
inmate-on-inmate violence d[id] not rise to the level necessary to
show deliberate indifference to a substantial risk of serious harm.”
Id. at 1235. We so held principally for two reasons.
First, we emphasized that Marbury’s only allegation proba-
tive of a “generalized risk of violence” was “his statement that he
personally witnessed fifteen inmate-on-inmate stabbings during his
time at St. Clair.” Id. at 1234. And importantly, we said, we lacked
evidence that could “place Marbury’s statement in context,” such
as “the total prison population,” “the sections of the prison in
which the attacks occurred,” or the “period of time [during which]
1 In response to one request, an officer allegedly responded, “You don’t enjoy
hanging out with the thugs, afraid you might get shanked!” and “I got the keys
to the city baby, you locked in.” Marbury I, 936 F.3d at 1231 (citation modi-
fied). In response to another, a captain allegedly responded, “[D]o you really
think I’ma act upon your requests, after you’ve filed complaints and requests
against me,” started laughing, then told Marbury to get himself a “shank.” Id.
at 1231–32. And in response to a third, Warden Estes, who is also a defendant
in this case, reportedly laughed at Marbury and told him to get a knife. Id. at
1232.
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 5 of 22
24-13513 Opinion of the Court 5
these incidents occurred.” Id. So far as we could tell, we said, Mar-
bury observed “fifteen stabbings . . . over the course of 6 years, for
a rate of 2.5 per year,” which we deemed insufficient to establish
the sort of generalized inmate-on-inmate violence that gives rise to
the requisite substantial risk of serious harm. Id.
Second, we emphasized that Marbury hadn’t—as successful
generalized-risk-of-violence plaintiffs had—“pointed to specific fea-
tures of [the] facility or its population rendering it particularly vio-
lent.” Id. at 1235. Examples of such “specific features,” we said,
might include “pervasive staffing and logistical issues rendering
prison officials unable to address near-constant violence, tensions
between different subsets of a prison population, and unique risks
posed by individual prisoners or groups of prisoners due to charac-
teristics like mental illness.” Id.
Accordingly, even though Marbury “faced some risk of as-
saults by fellow prisoners,” we held that “some risk” was “insuffi-
cient.” Id. at 1235. Because he hadn’t either provided “context” for
the incidents he observed or identified “specific features” making
St. Clair particularly violent, his deliberate-indifference claim
failed. Id. at 1234–35.
B
Marbury was transferred out of St. Clair sometime after he
sustained the 2016 attack. But in 2017, he was temporarily reas-
signed to St. Clair. The events following his reassignment gave rise
to this case. We recount those events in the light most favorable
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 6 of 22
6 Opinion of the Court 24-13513
to Marbury, as the nonmoving party at summary judgment below.
See Caldwell v. Warden, 748 F.3d 1090, 1098 (11th Cir. 2014).
Having been attacked once already in 2016, Marbury again
feared for his safety. See Marbury I, 936 F.3d at 1231–32. Accord-
ingly, he asked three prison officials to place him in segregation—
but to no avail. Marbury asked Warden Cedric Specks, who ad-
vised him that he should “obtain a knife or pay prison gang mem-
bers to protect him.” R&R, Aug. 12, 2024, at 5, Dkt. No. 87. He
also asked Captain Carla Graham, who dismissed his concerns be-
cause “the guy who stabbed [him] had been transferred” to a dif-
ferent prison. Id. at 5–6. And he mailed a written request to War-
den DeWayne Estes, who said he never received it. Id. at 6.
Denied the requested transfer, Marbury stayed in the gen-
eral prison population, where he alleges he saw several stabbings,
including one incident in which a supervisor confiscated a knife
from an inmate who was fighting with another prisoner, only to
give it back. Id. Marbury also claims that inmates occupied cells
to which they weren’t assigned. Id. And he says he felt unsafe in
the L/M Block, in which he was housed, “due to a lack of security
and ‘hostile living condition[s],’” id., and because guards were ab-
sent from the block for hours at a time, see Am. Compl. at 11, Dkt.
No. 9. Marbury claims that he reported these incidents and condi-
tions to prison staff.
Marbury’s fears turned out to be well-founded. On the
morning of January 9, 2018, an inmate entered his cell and hit him
on the head with a 20-pound dumbbell, knocking him unconscious.
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 7 of 22
24-13513 Opinion of the Court 7
Marbury sustained serious injuries, was transported to the emer-
gency room, and remained hospitalized for three days.
C
Just as Marbury had sued after the stabbing incident in 2016,
he brought this case following the attack he sustained in 2018.
Once again proceeding pro se, he filed a § 1983 action against War-
den Estes, Warden Specks, and Captain Graham (together, the
“Prison Officials”) in their individual capacities for violations of the
Eighth Amendment. He later filed an amended complaint clarify-
ing that he sought monetary, injunctive, and declaratory relief.
The magistrate judge recommended dismissal of all Mar-
bury’s claims except his Eighth Amendment claim based on the
theory that a generalized risk of violence pervaded St. Clair. The
district court accepted the magistrate judge’s recommendation.
The magistrate judge then ordered the Prison Officials to file
“special reports” addressing Marbury’s generalized-risk allega-
tions. 2 After the Prison Officials filed reports denying the charges,
2 Federal district courts in Alabama often use a “special report” proceeding to
process § 1983 claims brought by pro se prisoners. Chapman v. Dunn, 129 F.4th
1307, 1320 (11th Cir. 2025) (Jordan, J., concurring). In a special-report proceed-
ing, “soon after a pro se prisoner files a complaint under 42 U.S.C. § 1983,” the
magistrate judge orders the defendant to submit a report concerning the plain-
tiff’s allegations, which may be accompanied by affidavits or other evidence.
Id. Once the report is submitted, the magistrate judge or district court can
convert it sua sponte “into a motion for summary judgment by the defend-
ant.” Id.
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 8 of 22
8 Opinion of the Court 24-13513
the magistrate judge construed them as motions for summary
judgment and gave Marbury 21 days to respond.
Marbury filed several responses opposing summary judg-
ment and moved for leave to conduct additional discovery. 3 He
requested all incident reports regarding inmate-on-inmate assaults
involving the use of weapons and all administrative security poli-
cies at St. Clair. See Mot. for Additional Disc. at 1, Dkt. No. 51. For
the years 2010 to 2018, he also requested all incident reports regard-
ing inmate-on-inmate assaults “involving the use of weapons that
resulted in injury[] or death at St. Clair,” as well as “all employee[]
complaints regarding security hazards within St. Clair.” Id.
The magistrate judge recommended that the district court
(1) deny Marbury’s motion for additional discovery and (2) grant
the Prison Officials’ motions for summary judgment on the ground
that he had failed to show a substantial risk of serious harm. Find-
ing Marbury’s discovery requests “overbroad” and unlikely to “pre-
sent facts essential to opposing the [Prison Officials’] motion for
summary judgment,” the magistrate judge doubted that the infor-
mation Marbury sought was relevant. R&R, Jan. 27, 2022, at 12,
Dkt. No. 62 (citation modified). Moreover, the magistrate judge
3 Marbury initially filed a First Request for Production and Inspection of Doc-
uments, but the magistrate judge’s “Order for Special Report expressly in-
structed the parties that, other than initial disclosures, no additional discovery
would be allowed absent leave of court.” R&R, Jan. 27, 2022, at 11 n.9, Dkt.
No. 62. Accordingly, on the same day, Marbury filed a motion for leave to
conduct additional discovery.
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 9 of 22
24-13513 Opinion of the Court 9
reasoned, “[e]ven if the information ha[d] some relevance, Mar-
bury ha[d] not demonstrated how denial of his request would re-
sult in ‘substantial harm’ to his case.” Id. Over Marbury’s objec-
tions, the district court accepted the magistrate judge’s recommen-
dation and dismissed Marbury’s claims with prejudice.
Marbury appealed the dismissal, and this Court vacated the
district court’s judgment, holding that Marbury was entitled to
some discovery. See Marbury v. Warden, No. 22-10916, 2022 WL
17175549, at *4 (11th Cir. Nov. 23, 2022) (per curiam) (Marbury II).
We explained that despite Marbury’s “overbroad” discovery re-
quests,
the district court abused its discretion by denying
Marbury’s request outright instead of limiting the
scope of the discovery to avoid subjecting the Prison
Officials to unnecessary or burdensome discovery,
such as by limiting discovery of these documents to
only the previous year or the previous few years.
Id. at *3.
We disagreed with the district court that Marbury’s discov-
ery requests were “irrelevant” because “evidence regarding vio-
lence in the prison more generally was the exact kind of infor-
mation this Court ha[d] said would be necessary to sustain his
claim” in Marbury I. Id. Moreover, we observed that the magistrate
judge’s denial of discovery prevented Marbury from accessing in-
formation that would have allowed him to respond to the Prison
Officials’ summary-judgment motion because, as a pro se prisoner,
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 10 of 22
10 Opinion of the Court 24-13513
he simply “would not have access to this information without dis-
covery.” Id.
We therefore remanded the case to the district court, where
the magistrate judge ordered Marbury to file a renewed motion for
leave to conduct additional discovery. When Marbury did so, the
magistrate judge granted his renewed motion in part and denied it
in part. The magistrate judge ordered the Prison Officials to sup-
plement their special reports with their responses to Marbury’s dis-
covery requests and allowed them to amend their reports. The
Prison Officials duly produced a spreadsheet reflecting that there
had been 112 inmate-on-inmate assaults at St. Clair in 2016 and 84
such assaults in 2017, along with incident reports corresponding to
each attack.
The magistrate judge again construed the special reports as
motions for summary judgment. And once again, he recom-
mended that the district court grant summary judgment for the
Prison Officials on the ground that Marbury had failed to show he
faced a substantial risk of serious harm. The magistrate judge so
recommended for two principal reasons.
First, though the documents produced by the Prison Offi-
cials showed that “inmate-on-inmate assaults occurred at St. Clair,
with some resulting in serious injury which required outside med-
ical attention, and two deaths,” the magistrate judge believed that
Marbury hadn’t “put these numbers into context” as required by
Marbury I—in particular, by “taking into consideration the number
of inmates housed at St. Clair.” R&R, Aug. 12, 2024, at 10, Dkt.
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 11 of 22
24-13513 Opinion of the Court 11
No. 87. Nor did Marbury show, the magistrate judge continued,
that “an excessive number of the assaults” occurred in the cell block
where he was housed, as L/M Block statistics revealed only 23 as-
saults, 11.5% of the roughly 200 that occurred at the prison. Id. at
10–11.
Second, the magistrate judge found it problematic that Mar-
bury had failed to provide “factual support” for his assertion that
inmates changed their cells without authorization—details such as
“when these unauthorized cell changes occurred, the identity of
the inmates involved, which cells were involved, or whether these
unauthorized cell changes contributed in any way to the assault
against him on January 9, 2018.” Id. at 11.
The magistrate judge concluded from what he perceived to
be the lack of context and factual support that Marbury had failed
to establish a genuine dispute of material fact as to whether a gen-
eralized risk of violence pervaded St. Clair—and, therefore,
whether he faced a substantial risk of serious harm. Id. The mag-
istrate judge thus recommended that the district court grant sum-
mary judgment to the Prison Officials. Id. at 12. And over Mar-
bury’s objections, the district court again adopted the magistrate
judge’s recommendation and dismissed Marbury’s claims with
prejudice.
Marbury filed this appeal, marking his third trip—and this
particular case’s second trip—to this Court.
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 12 of 22
12 Opinion of the Court 24-13513
II
A prison official’s “deliberate[] indifferen[ce] to a substantial
risk of serious harm to an inmate who suffers injury” gives rise to
an Eighth Amendment violation. Lane v. Philbin, 835 F.3d 1302,
1307 (11th Cir. 2016). Though “[n]ot every injury suffered by one
inmate at the hands of another . . . translates into a constitutional
liability for prison officials responsible for the victim’s safety,”
“prison officials must take reasonable measures to guarantee the
safety of the inmates.” Marbury I, 936 F.3d at 1233 (citation modi-
fied). “To establish a § 1983 claim for deliberate indifference, a
plaintiff must show ‘(1) a substantial risk of serious harm; (2) the
defendants’ deliberate indifference to that risk; and (3) causation.’”
Id. (quoting Lane, 835 F.3d at 1307).
Because the district court dismissed Marbury’s deliberate-in-
difference claim at step one, so to speak—i.e., on the ground that
Marbury hadn’t shown a “substantial risk of serious harm”—we ad-
dress here only that threshold element. To show a substantial risk
of serious harm based on a “generalized risk of violence”—as Mar-
bury seeks to do—a plaintiff must meet a high bar. In particular,
he must show that the prison’s conditions were objectively “ex-
treme and posed an unreasonable risk of serious injury to his future
health or safety.” Id. (quoting Lane, 835 F.3d at 1307). And to make
that showing with respect to “serious inmate-on-inmate violence,”
he must demonstrate that such violence is “the norm or something
close to it,” such as “confinement in a prison where violence and
terror reign.” Id. at 1234 (quoting Purcell ex rel. Estate of Morgan v.
Toombs Cnty., 400 F.3d 1313, 1320, 1322 (11th Cir. 2005)). Neither
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 13 of 22
24-13513 Opinion of the Court 13
“a generalized awareness of risk” nor the mere presence of “occa-
sional, isolated attacks by one prisoner on another” will suffice. Id.
(citation modified).
Marbury contends that he has shown at least a genuine dis-
pute of material fact as to whether he faced a substantial risk of
serious harm at St. Clair. He points, of course, to the Prison Offi-
cials’ discovery responses, which reflect that there were 112 in-
mate-on-inmate assaults at the prison in 2016 and 84 such assaults
in 2017—an average of about nine and seven per month, respec-
tively, or about one every three or four days. Marbury also insists,
importantly, that he has satisfied Marbury I’s guidelines for estab-
lishing a generalized risk of violence because (1) the prison records
obtained in discovery provide sufficient “context” for his allega-
tions and (2) his sworn pro se pleadings identify “specific features”
that make St. Clair especially violent. Id. (“context”); id. at 1235
(“specific features”).
For the reasons that follow, we agree that Marbury has pro-
vided enough to survive summary judgment on the substantial-risk
issue. 4
4 “We review the district court’s grant of summary judgment de novo, viewing
all the evidence and drawing all reasonable inferences in the light most favor-
able to the non-moving party.” Marbury I, 936 F.3d at 1232 (citation modified).
“Summary judgment is warranted where the evidence in the record presents
no genuine issue of material fact and compels judgment as a matter of law in
favor of the moving party.” Id. (citation modified). “Where, as here, an in-
mate proceeded pro se in the district court, his summary judgment pleadings
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 14 of 22
14 Opinion of the Court 24-13513
A
As an initial matter, we hold that Marbury has provided the
requisite “context” to support his statistical allegations. In Marbury
I, we emphasized that although we had been presented with raw
data regarding the prevalence of inmate-on-inmate violence, we
lacked any evidence that would allow us to put those data in “con-
text.” Id. at 1234. In particular, we emphasized that we knew noth-
ing about “the total prison population,” “the sections of the prison
in which the attacks occurred,” or “period of time [during which]
these incidents occurred.” Id. Here, we have enough of that kind
of evidence.
For starters, we know St. Clair’s “total inmate population”
during the years in question. Publicly available government rec-
ords show that St. Clair housed 977 inmates as of December 2016
and 1,016 inmates as of December 2017—the two years leading up
to the attack on Marbury in January 2018. See Statistical Reports,
Ala. Dep’t of Corrs., https://doc.alabama.gov/StatReports.aspx. 5
are construed liberally and ‘specific facts’ alleged in his sworn complaint can
suffice to generate a genuine dispute of fact.” Id.
5 Because St. Clair’s inmate population is “not subject to reasonable dispute,”
we may take judicial notice of it. Fed. R. Evid. 201. Indeed, we’ve previously
taken judicial notice of similar facts published by state prison authorities. See,
e.g., Dimanche v. Brown, 783 F.3d 1204, 1213 n.1 (11th Cir. 2015) (taking judicial
notice of the numbers of colonels and prisons in Florida’s Department of Cor-
rections as published in the annual statistics report on the state’s Department
of Corrections’ website); Terrebonne v. Blackburn, 646 F.2d 997, 1000 n.4 (5th
Cir. June 1981) (taking judicial notice of Louisiana’s Board of Pardons’ pardon
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 15 of 22
24-13513 Opinion of the Court 15
Given the number of inmate-on-inmate assaults that occurred dur-
ing those years—again, 112 in 2016 and 84 in 2017—we can deduce
an average of one assault for every 10 inmates during the years in
question. Those numbers dwarf the assault statistics we deemed
insufficient in Purcell, 400 F.3d at 1322 n.21, and Harrison v. Culliver,
746 F.3d 1288, 1299–1300 (11th Cir. 2014), and we have little trou-
ble concluding that they create a genuine dispute of material fact
whether “violence was the norm or something close to it” at St.
Clair. Marbury I, 936 F.3d at 1234 (quoting Purcell, 400 F.3d at
1322). 6
Moreover, the incident reports disclosed in discovery con-
tain the other kinds of evidence that we said would help to contex-
tualize Marbury’s statistical allegations—namely, “the sections of
the prison” in which the incidents occurred and “over what period
or commutation rate as published in the state’s department of corrections’ an-
nual statistical reports). And judicial notice is particularly appropriate here
because, as we observed in Marbury II, pro se prisoners like Marbury have lim-
ited access to even publicly available information. See Marbury II, 2022 WL
17175549, at *3.
6 In Purcell, we held that “two to three pretty serious inmate fights over a pe-
riod of nine months” in a jail that housed more than 100 inmates didn’t estab-
lish “frequent and pervasive” instances of “serious inmate-on-inmate vio-
lence.” 400 F.3d at 1322 n.21. Controlling for the prison population, St. Clair
has at least two to four times as many violent incidents as in Purcell. In Harri-
son, we held that 33 incidents involving weapons over a period of 32 months
at a prison that housed between 830 and 990 inmates was insufficient to estab-
lish a substantial risk of serious harm. 746 F.3d at 1299–1300. Controlling for
the prison population, St. Clair had five to nine times as many violent incidents
in the relevant period as the prison in Harrison.
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 16 of 22
16 Opinion of the Court 24-13513
of time.” Id. In every respect, therefore, the record in this case is
far more extensive than that in Marbury I, in which Marbury’s lone
relevant allegation—that he had witnessed 15 stabbings—was un-
supported by any contextual evidence at all. Id.
The Prison Officials offer two responses to Marbury’s con-
textualized statistics. First, they emphasize that Marbury was an
unusual inmate in that he was part of a contingent of inmates who
were temporarily transferred from another prison and was housed
with them in the “L/M Block” at St. Clair. Accordingly, the Prison
Officials assert, we should consider only the statistics pertinent to
the L/M Block, rather than those about the total prison population.
And, their argument goes, the L/M Block’s inmate-assault statistics
more closely resemble those that we rejected in Purcell and Harri-
son. But in the particular circumstances of this case, and based on
the record as it exists before us, it’s not at all clear that being a trans-
fer in the L/M Block exposes an inmate to less violence than being
housed anywhere else. Assuming Marbury’s sworn allegations to
be true—which we must at this stage—St. Clair inmates roamed
unauthorized throughout the prison, weapons were prevalent, and
guards were often absent. Am. Compl. at 11–12, Dkt. No. 9. In
those circumstances, it stands to reason that the conditions existing
within St. Clair generally would pervade the entire facility and in-
fect the L/M Block, and that the prison-wide statistics would thus
have a bearing on the risks faced by L/M-Block inmates.
Second, the Prison Officials assert that statistics regarding in-
mate-on-inmate assaults in 2016 aren’t relevant because the present
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 17 of 22
24-13513 Opinion of the Court 17
case is about the conditions that Marbury faced when he returned
to St. Clair in 2017. Their argument is misguided for two reasons.
For one, even if we were to ignore the 2016 statistics, that would
leave 2017, in which there were 84 inmate-on-inmate assaults—or,
again, roughly seven per month and one every four days. For an-
other, we rejected so myopic a focus in Marbury II. There, we held
that the district court had erred in refusing as irrelevant Marbury’s
discovery requests for incident reports on inmate-on-inmate as-
saults from 2010 to 2018, and we emphasized that “evidence re-
garding violence in the prison more generally was the exact kind of
information this Court has said would be necessary to sustain [Mar-
bury’s] claim.” Marbury II, 2022 WL 17175549, at *3 (emphasis
added). We reasoned that the district court should have instead
“limit[ed] discovery . . . to only the previous year or the previous few
years.” Id. (emphasis added). And on remand, the district court
appropriately allowed discovery of those records for the years 2016
and 2017. Cf. Hale v. Tallapoosa Cnty., 50 F.3d 1579, 1583 (11th Cir.
1995) (looking to the prevalence of inmate-on-inmate violence dur-
ing the two years preceding the attack in question).
* * *
For these reasons, we hold that when placed in the appro-
priate “context,” the statistics that Marbury has presented—reflect-
ing almost 200 inmate-on-inmate assaults during the two years
leading up to the attack on him—establish a genuine factual dispute
as to whether he faced a substantial risk of serious harm at St. Clair.
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 18 of 22
18 Opinion of the Court 24-13513
B
In addition to evidence that allows us to evaluate Marbury’s
statistical evidence in “context,” Marbury’s sworn complaint and
accompanying affidavit point to “specific features” of St. Clair that
exacerbate the risks of inmate-on-inmate violence and that, accord-
ingly, reflect a genuine dispute of material fact as to whether he
faced a substantial risk of serious harm.
As relevant here, Marbury’s pro se pleadings contained the
following sworn allegations:
• Marbury “witness[ed] several stabbing incidents,” including
“one where a supervisor took a knife only to give it back to
the inmate he took it from to go stab[] the inmate he just got
into it with.” Am. Compl. at 11, Dkt. No. 9.
• Warden Specks suggested that Marbury “go get a knife or
pay the Crips, I.G.Ds, [or] Blood [gang] for some protec-
tion.” Id.
• “[A]t times the cubicle in [the] Block would be [un]occupied
with no officers on post for hours.” Id.
• “[T]here were inmates living in cells that w[ere] not assigned
to these cells by the administration at St. Clair. These in-
mates took the initiati[ve] to move without . . . authoriza-
tion.” Id. at 12.
On appeal, the Prison Officials have identified what they
take to be several problems with Marbury’s allegations. We’ll ad-
dress their objections in turn.
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 19 of 22
24-13513 Opinion of the Court 19
First, the Prison Officials assert that Marbury failed to point
to the right kind of “specific features”—“understaffing,” “over-
crowding,” lack of a “classification or segregation system,” or an
“inability to monitor [the] Block.” Supp. Br. of Appellees at 19. But
the requirement that an inmate point to “specific features” that
make a prison particularly dangerous isn’t a rigid box-checking ex-
ercise; there is no mandatory catalogue or hierarchy. Our en banc
decision in Marsh v. Butler County, 268 F.3d 1014 (11th Cir. 2001) (en
banc), abrogated on other grounds by Bell Atl. Corp. v. Twombly, 550
U.S. 554 (2007), makes that clear. There, we held that, if true, the
plaintiffs’ qualitative allegations regarding what Marbury I would
later call “specific features”—namely, that the prison’s locks didn’t
work, that inmates had ready access to weapons, and that no video
or audio surveillance system was in place—were sufficient to show
a substantial risk of serious harm. Id. at 1028. We emphasized that
the plaintiffs there weren’t required to provide, in addition, quan-
titative evidence that the prison had a “history of inmate assaults
with serious injuries.” Id. at 1034. Marbury has offered more—he
has both shown a history of inmate violence and pointed to several
features that make St. Clair particularly violent.
Second, the Prison Officials argue that Marbury’s “limited
evidence” fails to sufficiently support his allegations of “unauthor-
ized cell movements and weapons availability.” Supp. Br. of Ap-
pellees at 19. In short, we disagree. With respect to inmates mov-
ing around the facility, the Prison Officials complain that Marbury
didn’t provide evidence of “when these unauthorized cell changes
USCA11 Case: 24-13513 Document: 44-1 Date Filed: 07/07/2026 Page: 20 of 22
20 Opinion of the Court 24-13513
occurred, the identity of the inmates involved, which cells were in-
volved, or whether these unauthorized cell changes contributed in
any way to the assault against him.” Id. at 20 (quoting R&R, Aug.
12, 2024, at 11, Dkt. No. 87). We see two problems. One, Marbury
needn’t go into such detail: This is an ordinary civil case, not a
fraud claim of the sort subject to heightened who-where-when
pleading standards. See Fed. R. Civ. P. 9(b). Moreover, and in any
event, Marbury was proceeding pro se in the district court, so we
must accept his sworn allegations as true at summary judgment
and construe his pleadings liberally. Marbury I, 936 F.3d at 1232.
Two, Marbury’s allegation is supported by the incident reports in
the record, several of which show that an inmate assigned to one
cell block was involved in an assault in a different cell block. See,
e.g., Resp. to Disc. Reqs. at 207, Dkt. No. 84–1; id. at 239; id. at 285
(reporting that an inmate involved in an incident “was down in H-
Dorm to miss all the dangers” of his assigned cell block). 7
The Prison Officials also downplay Marbury’s allegations re-
garding the ready availability of weapons inside the prison. They
contend that the “few stabbings” in the record and Specks’s advice
that Marbury should “get a knife” to protect himself aren’t suffi-
cient to “give rise to a reasonable inference of widespread weapons
availability.” Supp. Br. of Appellees at 21 (emphasis in original).
7 Cf. Boykins, 696 F. Supp. 3d at 1069–71 (noting Warden Estes’s testimony that
St. Clair’s policy for keeping track of inmate movement was unenforced and
that “probably more than 50 percent” of incidents at the prison were “caus