United States of America v. Erik Matthew Harris

PRECEDENTIAL
UNITED STATES COURT OF APPEALS
FOR THE THIRD CIRCUIT
_______________
No. 21-3031
_______________
UNITED STATES OF AMERICA
v.
ERIK MATTHEW HARRIS,
Appellant
_______________
On Appeal from the United States District Court
for the Western District of Pennsylvania
(D.C. No. 2:19-cr-00313-001)
District Judge: Hon. Marilyn J. Horan
_______________
Argued: December 9, 2024
Before: KRAUSE, BIBAS, and AMBRO, Circuit Judges
(Filed: July 14, 2025)
Renee Pietropaolo [ARGUED]
FEDERAL PUBLIC DEFENDER’S OFFICE

-- 1 of 48 --

2
1001 Liberty Avenue
1500 Liberty Center
Pittsburgh, PA 15222
Counsel for Appellant
Laura S. Irwin
UNITED STATES ATTORNEY’S OFFICE
700 Grant Street
Suite 4000
Pittsburgh, PA 15219
Andrew C. Noll [ARGUED]
UNITED STATES DEPARTMENT OF JUSTICE
CRIMINAL DIVISION, APPELLATE SECTION
950 Pennsylvania Avenue NW
Room 1252
Washington, DC 20530
Counsel for Appellee
_______________
OPINION OF THE COURT
_______________
BIBAS, Circuit Judge.
Guns and drugs can be a lethal cocktail. So Congress passed
18 U.S.C. § 922(g)(3), barring illegal drug users from having
guns until they stop using. Erik Harris, a frequent marijuana
smoker, bought guns anyway. He was convicted of possessing
them and of lying about his drug use to get them. Now he chal-
lenges those convictions, claiming that the gun ban for illegal

-- 2 of 48 --

3
drug users violates the Second Amendment and is unconstitu-
tionally vague.
Today, we hold that history and tradition justify
§ 922(g)(3)’s restrictions on those who pose a special danger of
misusing firearms because they frequently use drugs. But we
lack enough facts to tell whether the law’s restrictions are con-
stitutional as applied to Harris. Still, § 922(g)(3) is not vague;
it warned Harris that he could not possess guns while routinely
smoking marijuana. So we will affirm in part, vacate in part,
and remand for the District Court to find facts needed to apply
the Second Amendment law laid out here.
I. WHILE SMOKING MARIJUANA REGULARLY,
HARRIS BUYS THREE GUNS
When Erik Harris was 21, he bought his first pistol. Before
buying the gun, Harris filled out a federal form that asked if he
was “an unlawful user of or addicted to marijuana.” 2 App.
199. He checked “no.” Eleven days later, he went back to the
same dealer to buy a second pistol. Again, he filled out the
same form. And again, he checked “no.”
Five days later, he went out partying with one of his new
guns. He got “really drunk” and high and, in the revelry, lost
his new gun. 3 App. 34. The next morning, he reported it sto-
len. Then he went back to the same dealer to buy a third pistol
as a replacement. Once again, he filled out the form. And even
though he had smoked marijuana the night before, he once
again checked “no” to being an unlawful user.
When Harris’s missing gun turned up in a felon’s hands,
officers called Harris in for questioning. There, he admitted

-- 3 of 48 --

4
that he smoked marijuana regularly, including earlier that same
day. But throughout the interview, he gave different estimates
of how often he had smoked in the past year. And he did not
say how much or how often he had smoked in the weeks lead-
ing up to and during his possession of the three guns.
When police asked him if, on the federal form, he had
answered honestly about his marijuana use, he hedged that it
“depends which way you look at it.” 3 App. 54. But he con-
ceded that he “didn’t answer honestly, for the most part” on the
form. 3 App. 58. He acknowledged being an “unlawful user”
of marijuana “because I do use it today.” 3 App. 53.
The government charged Harris with three counts under 18
U.S.C. § 922(g)(3) for possessing each gun as an “unlawful
[drug] user” and three counts under § 922(a)(6) for lying to buy
each one. 2 App. 25. Harris moved to dismiss all counts. He
argued that § 922(g)(3) violates the Second Amendment as ap-
plied to him. He also argued that the phrase “unlawful user” is
unconstitutionally vague, invalidating both § 922(g)(3) (which
bars “unlawful user[s]” from having guns) and § 922(a)(6)
(which bars lying about being an unlawful user).
The District Court denied Harris’s motion. It made no spe-
cific finding about how much or how often Harris was smoking
in the weeks around his gun possession. But it concluded that
§ 922(g)(3) was constitutional as applied to Harris, using
means-end scrutiny under Binderup v. Attorney General, 836
F.3d 336, 353, 356 (3d Cir. 2016) (en banc). Then Harris
pleaded guilty to all six counts, preserving his right to appeal
the issues raised in his motion to dismiss. We review the

-- 4 of 48 --

5
District Court’s denial de novo. United States v. Gonzalez, 905
F.3d 165, 190 (3d Cir. 2018).
II. TWO ANALOGUES JUSTIFY § 922(g)(3)’S
RESTRICTIONS FOR SOME DRUG USERS
Section 922(g)(3) bans possession of a gun by anyone “who
is an unlawful user of or addicted to any controlled substance.”
Harris claims that this ban violates his Second Amendment
rights. To assess that claim, we use a two-step test focused on
the Amendment’s “text and historical understanding.” N.Y.
State Rifle & Pistol Ass’n v. Bruen, 597 U.S. 1, 26 (2022). First,
we decide whether the Amendment’s text covers his conduct.
Id. at 17. If it does, the government can justify disarming him
only if doing so is “consistent with this Nation’s historical tra-
dition of firearm regulation.” Id.
At step one, the Second Amendment presumptively pro-
tects Harris’s conduct. Drug users who are adult citizens are
among “the people” who fall within its scope. Range v. Att’y
Gen., 124 F.4th 218, 226–28 (3d Cir. 2024) (en banc). And
§ 922(g)(3) regulates “quintessential Second Amendment con-
duct: possessing a handgun.” United States v. Moore, 111 F.4th
266, 269 (3d Cir. 2024).
So our inquiry turns on the second step: whether disarming
Harris is “consistent with the principles that underpin our reg-
ulatory tradition.” United States v. Rahimi, 602 U.S. 680, 692
(2024). Modern laws pass this test if they are “ ‘relevantly sim-
ilar’ to laws that our tradition is understood to permit,” espe-
cially in “[w]hy and how [they] burden[ ] the right.” Id. (quot-
ing Bruen, 597 U.S. at 29).

-- 5 of 48 --

6
Though our Second Amendment law looks to history and
tradition, it is not “trapped in amber.” Id. at 691. The Amend-
ment “permits more than just those regulations identical to
ones that could be found in 1791.” Id. at 692. We should not
“assume[ ] that founding-era legislatures maximally exercised
their power to regulate” and thus that every novel regulation is
unconstitutional. Id. at 739–40 (Barrett, J., concurring). Mod-
ern regulations must rest on historical “principles” but need not
squeeze into narrower historical “mold[s].” Id. at 692 (major-
ity), 740 (Barrett, J., concurring). This means that the govern-
ment need identify only a “historical analogue,” not a “histori-
cal twin.” Id. at 701 (majority) (quoting Bruen, 597 U.S. at 30).
And the analogy turns on similarity in principle, not specific
facts: A historical law is a fitting analogue for a modern one if
it burdens Second Amendment rights for comparable reasons
(the “why”) using comparable means (the “how”). Id. at 692.
The most obviously applicable historical tradition here
would be one regulating gun possession by marijuana users.
Yet no Founding-era law disarmed them. That is no surprise.
Despite speculation that some Founders smoked hemp, it was
mainly a source of cloth, paper, and rope, not a drug. See Mar-
tin Booth, Cannabis: A History 33–37 (2003).
But the government identifies historical cousins to
§ 922(g)(3) that it says justify its restrictions: regulations on the
dangerously drunk and dangerously mentally ill. We survey
these analogues below and conclude that they support
§ 922(g)(3)’s constitutionality as applied to those whose drug
use would likely cause them to pose a physical danger to others
if armed.

-- 6 of 48 --

7
A. The Founding Generation incapacitated drunks who
posed a risk of danger to others
After marijuana, the next most intuitive analogue to the
modern mind is alcohol. Drinking lots of alcohol was a normal
part of colonial life. Mark Edward Lender & James Kirby Mar-
tin, Drinking in America: A History 9–14 (1987). But the
Founders also understood that drinking could provoke people
to act dangerously. In England, drunkenness was widely decried
as contributing to crime and violence. See Dana Rabin, Drunk-
enness and Responsibility for Crime in the Eighteenth Century,
44 J. Brit. Studies 457, 459–66 (2005). So in 1606, England
banned public drunkenness, declaring it “the roote and
foundacion of many other enormious Synnes, as Bloodshed
Stabbinge Murder … and such lyke.” 4 Jac. 1, c. 5 (1606); see
also 4 William Blackstone, Commentaries *64 (discussing the
law). And justices of the peace could require twice-convicted
drunks to post sureties for their good behavior; drunks who did
not comply could be jailed. Michael Dalton, The Country Jus-
tice: The Practice, Duty and Power of the Justices of the Peace
289 (London, Henry Lintot 1746).
When the Founders crossed the Atlantic, they carried these
concerns with them. Dr. Benjamin Rush, a signer of the Dec-
laration of Independence, delegate to the Continental Con-
gress, and preeminent Founding-era medical authority, noted
that intoxication breeds crime, including “[f]ighting,” “[b]ur-
glary,” and “[m]urder.” Benjamin Rush, An Inquiry into the
Effects of Ardent Spirits upon the Human Body and Mind 2 (8th
ed., Boston, James Loring 1823). Likewise, he viewed “crimes
and infamy … [as the] usual consequences of the intemperate
use of ardent spirits.” Id. at 13.

-- 7 of 48 --

8
States recognized the danger of mixing alcohol with guns.
An early Rhode Island law banned firing guns at night and in
taverns, and a New York law barred shooting around New
Year’s Eve to prevent damage caused by combining alcohol
with firearms. See Acts & Laws of the English Colony of
Rhode-Island & Providence Plantations 120 (Newport, Hall
1767); Act of Feb. 16, 1771, ch. 1501, reprinted in 5 The Co-
lonial Laws of New York from the Year 1664 to the Revolution
244, 244–46 (Albany, James B. Lyon 1894).
Plus, early legislatures authorized constables to confine
drunks who posed a risk to others until they sobered up. See,
e.g., General Laws and Liberties of the Massachusetts Colony
81 (1672) (authorizing officers to imprison any drunk who was
“abus[iv]e to” or was “striking” others); Grants, Concessions,
and Original Constitutions of the Province of New-Jersey: The
Acts Passed during the Proprietary Governments, and Other
Material Transactions before the Surrender Thereof to Queen
Anne 107 (1753) (ordering that drunks who “are unruly and
disturbers of the Peace, shall be put in the Stocks, until they are
Sober”).
Some jurisdictions even locked up anyone found drunk in
public. Act of June 18, 1807, reprinted in Laws of the State of
New-Hampshire, Passed from December Session, 1805, to
June Session, 1810, Inclusive 74 (Concord, N.H., Isaac Hill
1811) (empowering officers to “arrest any … common drunk-
ards” found at night and keep them in jail “until the following
day”); Act of Sept. 17, 1807, reprinted in Compend of the Acts
of Indiana, From the Year Eighteen Hundred and Seven Until
That of Eighteen Hundred and Fourteen, Both Inclusive 54–

-- 8 of 48 --

9
55, 91 (Vincennes, Ind., Elihu Stout 1817) (ordering justices of
the peace to imprison noisy drunks for up to “48 hours”).
Others enacted surety regimes empowering magistrates to
make drunkards give security for peace and good behavior or
be imprisoned. Acts and Laws of his Majesties Colony of
Rhode-Island, and Providence-Plantations in America 11
(Boston, John Allen 1719); An Act Against Breaking the
Peace, reprinted in Acts and Laws of the State of Connecticut,
in America 189 (Hartford, Elisha Babcock 1786); The Public
Laws of the State of South-Carolina App. II at 26 (Philadel-
phia, Aitken & Son 1790); Act of December 26, 1792, in Di-
gest of the Laws of Virginia Which are A Permanent Character
and General Operation 756 n.2 (Richmond, Smith & Palmer
1841); Act of Dec. 16, 1812, in A Digest of the Laws of the
Corporation of the City of Washington to the First of June,
1823, at 141 (Washington, D.C., James Wilson 1823) (requir-
ing those “found … drunk in or about the streets” to “enter into
security for good behaviour for a reasonable time” or be sen-
tenced to 90 days’ labor); see also 1 Laws of the State of Dela-
ware 173–74 (New Castle, Del., Samuel & John Adams 1797)
(punishing drunkenness and requiring drunks who abused arrest-
ing officers to be “bound to his or her good behaviour” as
“breaker[s] of the peace”); A Digest of the Laws of Maryland
206 (Baltimore, Thomas Herty ed., 1799) (punishing drunken-
ness and requiring any drunkard who “revil[ed]” arresting offic-
ers to “give security … for his good behaviour for three
months[ ] or suffer one month’s imprisonment without bail”).
Like the surety laws that the Court relied on in Rahimi,
these regimes were a “form of preventive justice”: Drunks had
to promise not to break the peace, lest they be locked up and

-- 9 of 48 --

10
thus disarmed. Rahimi, 602 U.S. at 695 (internal quotation
marks omitted).
B. The Founding Generation likewise incapacitated the
mentally ill who risked endangering others
After alcohol, the next most fitting historical analogy to
marijuana is how the Founders thought about the danger posed
by the mentally ill. “ Obviously, mental illness and drug use
are not the same thing.” United States v. Veasley, 98 F.4th 906,
912 (8th Cir. 2024). But in dealing with a new social problem
like habitual marijuana use, as Judge Stras has explained, “we
cannot look at history through a pinhole.” Id. (also providing
many of the sources discussed below). Instead, we must
broaden our view by looking at how the Founding generation
addressed all analogous problems.
And the Founding Generation often analogized intoxication
to mental illness, sharing our modern intuition that “their behav-
ioral effects overlap.” Id. They understood “habitual drinking”
as, in part, a form of “blameless insanity.” Erik Fisher, The
Urge: Our History of Addiction 47 (2022). Dr. Rush described
drunkenness as “a temporary fit of madness.” Rush at 6. Like-
wise, one of Dr. Rush’s patients, a chronic drug user, was diag-
nosed as suffering “[i]nsanity from the use of Opium.” Eliza-
beth Kelly Gray, Habit Forming: Drug Addiction in America,
1776–1914, at 20 (2023). Medical observers started seeing exces-
sive drinking as a “significant” trigger of “madness.” Mary
Ann Jimenez, Changing Faces of Madness 72 (1987). Thomas
Cooley’s influential treatise later drew the same comparison.
Thomas M. Cooley, A Treatise on the Constitutional Limita-
tions Which Rest upon the Legislative Power of the States of

-- 10 of 48 --

11
the American Union *599 n.1 (Boston, Little, Brown & Co.
1868) (“Drunkenness is regarded as temporary insanity.”).
The analogy fit because the Founders understood mental
illness as “a transitory condition, just like intoxication,” that
impaired one’s mental faculties temporarily. Veasley, 98 F.4th
at 913. Those who suffered from bouts of mental illness were
called “lunatic[s],” drawn from the Latin word for the moon,
on the belief that they “ha[d] lucid intervals, sometimes enjoy-
ing [their] senses, and sometimes not, and that frequently depend-
ing on the change of the moon.” 1 Blackstone at *304. (Because
we focus on the history, we use the historical terms even
though they offend modern ears.) Lunatics intermittently “lost
the use of [their] reason” and regained it. Id.; accord Anthony
Highmore, A Treatise on the Law of Idiocy and Lunacy 3,
104–05 (London, R. Wilks 1807); Lunatic, in 3 A New and
Complete Dictionary of Arts and Sciences 1951 (London,
Society of Gentlemen 1754). “[T]he law always imagine[d]
that these accidental misfortunes [of mental illness] may be
removed ….” 1 Blackstone at *305.
Still, the Founders understood that some lunatics posed a
risk of endangering others because of their mental state. When
a lunatic posed no danger to society, he remained free, retain-
ing an ordinary citizen’s rights and responsibilities. Veasley,
98 F.4th at 913; see also Highmore at 128–29 (stating that a
will that a lunatic drew up while lucid was valid). But when
officials determined that a person might be in a state of lunacy
that would pose a danger to society, the lunatic temporarily lost
his liberty. He was locked up in a jail, hospital, or asylum until
the threat he posed abated. In eighteenth-century England, jus-
tices of the peace could lock up those “who by Lunacy, or

-- 11 of 48 --

12
otherwise, are furiously mad, or are so far disordered in their
Senses that they may be dangerous to be permitted to go
abroad.” Justices Commitment Act of 1743, 17 Geo. 2, c. 5,
§ 20 (Eng.) (emphasis added). These restrictions were usually
temporary, lasting “only so long as such lunacy or disorder
shall continue, and no longer.” Henry Care & William Nelson,
English Liberties, or the Free-Born Subject’s Inheritance 329
(6th ed., Providence, John Carter 1774).
The colonists brought these English practices with them
across the Atlantic. Philadelphians who were both mentally ill
and dangerous “were confined in barred cells in the basement”
of a hospital; “particularly violent individuals” were “restrained
… using a ‘straitwaistcoat’ or ‘mad shirt,’ or heavy arm and
leg chains.” Lynn Gamwell & Nancy Tomes, Madness in
America 20 (1995). New York provided that the “furiously
mad” could be “kept safely locked up” and “chained.” An Act
for Apprehending and Punishing Disorderly Persons, c.31
(1788), reprinted in 2 Laws of the State of New York Passed at
the Sessions of the Legislature Held in the Years 1785, 1786,
1787 and 1788, Inclusive 643, 645 (Albany, Weed, Parsons &
Co. 1886). Connecticut obligated local authorities to lock up
temporary lunatics who might “endanger[ ] [others] in person
or estate.” An Act for Relieving and Ordering of Idiots, Impo-
tent, Distracted, and Idle Persons § 18 (1793), in 1 The Public
Statute Laws of the State of Connecticut 382, 386 (Hartford,
Hudson & Goodwin 1808). And Massachusetts authorized jus-
tices to confine any person “so furiously mad as to render [him]
dangerous to the peace and safety of the good people.” Stat.
1797, c. 62, § 3, reprinted in 2 Compendium and Digest of the

-- 12 of 48 --

13
Laws of Massachusetts 688 (Boston, Thomas B. Wait & Co.
1810).
To justify incapacitating someone, these officials had to
predict whether he would become dangerous in periods of
lunacy. They did not have to wait until he had harmed someone
else. See Jimenez at 91–92 (noting that Boston’s “maniac
house” confined not only the “turbulent and almost ungovern-
able,” but also some who “were only ‘periodically’ insane” and
even some who “were calm and not violent” because they
“were viewed as potentially dangerous”). Rather, they had “a
lot of discretion” to discern when someone posed enough of a
threat to public safety to warrant confinement. Veasley, 98
F.4th at 914. For instance, one manual for justices of the peace
explained that “[a]ny person” could “confin[e]” a lunatic “as is
proper in such circumstances.” Lunatics, in James Parker, Con-
ductor Generalis: Or the Office, Duty and Authority of Justices
of the Peace, High-Sheriffs, Under-Sheriffs, Coroners, Consta-
bles, Goalers [sic], Jury-Men, and Overseers of the Poor 290,
291 (New York, John Patterson 1788).
And these same officials had to make judgment calls to dis-
cern when a person no longer needed to be confined. Officials
could not possibly monitor a person’s mental state moment by
moment. So people who were in a state of lunacy were “viewed
as potentially dangerous,” even if they were periodically lucid.
Jimenez at 91–92. They could thus remain locked up until an
official determined that the threat they posed had fully abated.
See id. Those judgment calls were not exact science. Justices
of the peace were typically laymen, trained neither in law nor
in medicine. Chester H. Smith, The Justice of the Peace System
in the United States, 15 Calif. L. Rev. 118, 127 (1927).

-- 13 of 48 --

14
Temporary imprisonment required temporary disarma-
ment. Those who were confined could not bring their guns with
them into confinement. Veasley, 98 F.4th at 913 (collecting
sources). So there was a longstanding, widespread tradition of
disarming people whose mental illnesses caused them to lose
their senses temporarily and pose a risk to others.
C. History and tradition thus support § 922(g)(3)’s con-
stitutionality as applied to drug users who would
pose a risk to others if armed
“Taken together,” these laws “confirm what common sense
suggests”: Someone who regularly uses mind-altering sub-
stances that make him a “credible threat to the physical safety
of others with a gun” may be disarmed temporarily until he
stops using drugs. Rahimi, 602 U.S. at 694, 698.
And that is precisely what § 922(g)(3) does. It bars gun pos-
session by anyone “who is an unlawful user of or addicted to
any controlled substance.” Controlled substances include
Schedule I drugs like heroin, LSD, and marijuana. 21 U.S.C.
§ 812, sched. I. An “addict” is anyone “who habitually uses
any narcotic drug so as to endanger the public morals, health,
safety, or welfare, or who is so far addicted to the use of nar-
cotic drugs as to have lost the power of self-control with refer-
ence to his addiction.” 21 U.S.C. § 802(1). And a “user” is any-
one who has “engaged in regular use [of drugs] over a period
of time proximate to or contemporaneous with the possession
of the firearm.” United States v. Augustin, 376 F.3d 135, 139
(3d Cir. 2004). So by its own terms, § 922(g)(3) temporarily
bars anyone who often uses drugs from possessing a gun
shortly before, during, or after using drugs.

-- 14 of 48 --

15
That restriction is well-grounded in history. By taking guns
out of the hands of frequent drug users, § 922(g)(3) addresses
a problem comparable to the one posed by the dangerously
mentally ill and dangerous drunks: a risk of danger to the pub-
lic due to an altered mental state. And its temporary restriction
on gun rights is analogous to these historical restrictions as
well. Of course, § 922(g)(3) “is by no means identical to” these
historical precursors. Id. “[B]ut it does not need to be”; its tem-
porary restriction on drug users who would pose a risk of dan-
ger with a gun in their hands “fits neatly” within this historical
tradition. Id.
First, the historical laws targeted a problem comparable to
the one that § 922(g)(3) addresses as applied to drug users who
would risk danger to others if armed. Both the drunkenness and
lunacy laws temporarily incapacitated people based on the
judgment that their impaired mental state posed a risk to others.
The lunacy laws authorized officials to confine lunatics based
on an individualized finding “that there would be some risk of
‘mischief’ without it.” United States v. Cooper, 127 F.4th
1092, 1096 (8th Cir. 2025) (emphasis added) (quoting Veasley,
98 F.4th at 914 (quoting Daniel Davis, A Practical Treatise
upon the Authority and Duty of Justices of the Peace in Crim-
inal Prosecutions 41 (Boston, Hilliard, Gray, Little, & Wilkins
2d ed. 1828))). This meant that someone could be deemed
“dangerous” and so needing confinement before he had
harmed or threatened anyone. Officials did not need to wait for
the danger to materialize. But such findings were still always
based on an “individualized assessment” rather than a categor-
ical judgment. Id.

-- 15 of 48 --

16
Likewise, Founding-era legislatures often required drunks
to post bonds for their good behavior or face imprisonment,
based on the judgment that drunks posed a risk to public peace
and the safety of others. See, e.g., Acts and Laws of the State of
Connecticut at 189; The Public Laws of the State of South
-Carolina App. II at 26. In this way, both the drunkenness and
lunacy laws operated like the surety and going-armed laws that
the Supreme Court blessed in Rahimi: They “permit[ted] the
disarmament of individuals who pose a credible threat to the
physical safety of others,” thereby “providing a mechanism for
preventing violence before it occurred.” Rahimi, 602 U.S. at
693, 697. And they both did so based on a risk assessment: Is
someone likely to pose a danger to others because of his im-
paired mental state?
Second, the historical laws imposed a comparable, indeed
greater, burden on gun rights. The lunacy laws authorized mag-
istrates to lock up in jails, hospitals, or asylums mentally ill
people who they found posed a risk to others. So did Massa-
chusetts’s and New Jersey’s drunkenness laws and many
states’ surety laws, which authorized locking up drunks who
broke their promises to stay dry. Imprisonment necessarily
involved disarmament. Veasley, 98 F.4th at 913 (collecting
sources). And “if imprisonment was permissible” to protect
“the physical safety of others, then the lesser restriction of tem-
porary disarmament that [§ 922(g)(3)] imposes is also permis-
sible.” Rahimi, 602 U.S. at 699.
Plus, § 922(g)(3)’s burden is temporary. It forbids gun pos-
session only as long as someone is using drugs regularly and
so “likely poses an increased risk of physical danger to others
if armed.” Pitsilides v. Barr, 128 F.4th 203, 212 (3d Cir. 2025)

-- 16 of 48 --

17
(internal quotation marks omitted). To be sure, drug users who
flout the ban will face a felony conviction, and felons can be
disarmed. See 18 U.S.C. § 922(g)(1). But as we have recog-
nized, disarmed felons may bring declaratory judgment actions
or petition the Attorney General to get their rights back. See
Range, 124 F.4th at 232; 18 U.S.C. § 925(c). Section
922(g)(3)’s “limited duration” thus tracks the historical restrictions
on lunatics or drunks, which were also temporary and ceased
once someone regained his senses or sobered up. Rahimi, 602
U.S. at 699.
Our partially dissenting colleague parts ways with us, con-
tending that a test focused on risk of danger impermissibly
lowers the bar. According to our dissenting colleague, some-
one must be “plainly” dangerous to be disarmed. Dissent at 6.
Anything else flouts the history and precedent that binds us, he
says. Yet it is the other way around. To start, the history clearly
shows that officials did not need to wait to act until a drunkard
or mentally ill person had harmed another. Rather, as our sister
circuit has acknowledged, officials determined whether to con-
fine a “lunatic” based on whether there would be “some risk of
mischief without it.” Cooper, 127 F.4th at 1096 (emphasis
added) (internal quotation marks omitted). The same goes for
drunks. True, as our dissenting colleague points out, a few laws
let officials confine drunks only after they acted abusively.
Dissent at 4–5. But that plucks out a handful of the dozen that
we cite, obscuring historical reality. See Bruen, 597 U.S. at 65–
66 (cautioning against cherry-picking history). Many other
laws deemed mere drunkenness sufficient to justify temporary
disarmament. See, e.g., Acts and Laws of the State of Connect-
icut at 189 (authorizing officials to make any “drunkard[ ]” post

-- 17 of 48 --

18
surety for good behavior or be imprisoned); The Public Laws
of the State of South-Carolina, App. II at 26 (same); Digest of
the Laws of Virginia Which are A Permanent Character and
General Operation at 756 n.2 (same); A Digest of the Laws of
the Corporation of the City of Washington at 141 (same); Acts
and Laws of the State of New-Hampshire at 74 (authorizing of-
ficials to temporarily confine drunks); Compend of the Acts of
Indiana at 54–55 (same). Even the Connecticut surety regime,
which our dissenting colleague says disarmed only those who
“terrif[ied] or disquiet[ed]” others, actually authorized officials
to require all “drunkards” to post surety for good behavior.
Laws of the State of Connecticut at 189; see Dissent at 4. Indeed,
at the Founding, the consensus was that surety laws extended
to all “common drunkards,” not just those who acted abusively.
4 Blackstone at *256; Parker at 348; Eliphalet Ladd, Burn’s
Abridgement, Or The American Justice 405–406 (2d ed. 1792).
So in this respect, we must part ways with the dissent on the
proper principle to extract from the history: Disarmament
based on danger was always based on a predictive judgment of
danger, not certainty.
Our colleague equally misses the mark when he claims that
we dodge binding precedent. True, Rahimi concluded that
someone found to “pose[ ] a clear threat of physical violence to
another” could be disarmed. Rahimi, 602 U.S. at 698. But
Rahimi did not exhaustively catalogue when someone could be
disarmed; it answered only that narrower question presented.
Id. at 684–85 (considering as-applied challenge to § 922(g)(8),
which disarms those found to be a “credible threat to the phys-
ical safety of [an] intimate partner” or child). Its holding
merely parroted the requirements of § 922(g)(8) itself. And it

-- 18 of 48 --

19
said no more and no less about the propriety of other restrictions.
See id. at 698. Indeed, Rahimi expressly declined to “undertake
an exhaustive historical analysis … of the full scope of the Sec-
ond Amendment.” Id. at 702 (internal quotation marks omit-
ted).
The dissent similarly errs in analyzing Pitsilides. There, we
considered only an as-applied challenge to § 922(g)(1), the
felon-in-possession law. And we held, based on the history we
recounted in Range, that the Second Amendment permits dis-
armament, “at a minimum,” when an individual “present[s] a
special danger of misusing firearms,” meaning he “would
likely pose a physical danger to others if armed.” Pitsilides,
128 F.4th at 210 (cleaned up). We said nothing about what
might justify a different regulation, such as § 922(g)(3), which
targets a different problem through different means.
Of course, the history justifying § 922(g)(8) or § 922(g)(1)
in some applications may be relevant to any Second Amend-
ment analysis. But it is only a slice of the history that reveals
the “principles that underpin our regulatory tradition.” Rahimi,
602 U.S. at 692. We must not blindly defer to historical princi-
ples extracted from partial surveys of history relevant to other
regulations.
Even so, we discern a similar principle from the relevant
history: Drug users can be disarmed based on the likelihood
that they will physically harm others if armed. But here is the
distinction: To assess that risk, judges need not wait until a
drug user has harmed or threatened another. They may decide
whether a drug user “would likely pose a physical danger to
others if armed” based on the nature of someone’s drug use and

-- 19 of 48 --

20
the risk that it will impair his ability to handle guns safely.
Pitsilides, 128 F.4th at 210 (cleaned up); see also id. at 212
(noting that conduct can be relevant if it bears on whether
someone “likely poses an increased risk of physical danger to
others if armed”). In divining this rule, we do not flout prece-
dent but abide by it. Bruen tells us to follow the history where
it leads. See Bruen, 597 U.S. at 17. That is all we do here.
Finally, the dissent protests the consequences, worried that
our holding will disarm even his hypothetical “hunters in a
duck blind.” Dissent at 13. But a buzzed brain with a loaded
gun sounds like a misfire waiting to happen—the exact risk
that our historical tradition suggests justifies disarmament.
*****
In sum, § 922(g)(3) temporarily and constitutionally restricts
the gun rights of drug users only as long as they “present a spe-
cial danger of misusing firearms.” Pitsilides, 128 F.4th at 211
(cleaned up); see also United States v. Daniels, 124 F.4th 967,
978–79 (5th Cir. 2025) (leaving open whether § 922(g)(3) is
constitutional as applied to some marijuana users); Cooper,
127 F.4th at 1098 (same).
III. ON REMAND, THE DISTRICT COURT MUST FIND
MORE FACTS TO DECIDE WHETHER § 922(g)(3)
IS CONSTITUTIONAL AS APPLIED TO HARRIS
The District Court let the government prosecute Harris
under § 922(g)(3) without finding that Harris’s frequent mari-
juana use increased the risk that he could not handle guns
safely. We do not fault the District Court for failing to make
this finding. When Harris moved to dismiss the indictment, the

-- 20 of 48 --

21
District Court was bound to apply the means-end scrutiny dic-
tated by Binderup. But then the Supreme Court decided Bruen,
“effect[ing] a sea change in Second Amendment law” and ab-
rogating that decision. Pitsilides, 128 F.4th at 208. Now “his-
tory and tradition,” not “legislative interest balancing,” dictates
whether a law comports with the Second Amendment. Bruen,
597 U.S. at 22, 26.
The District Court had no chance to take the first crack at
whether § 922(g)(3) is constitutional as applied to Harris under
this proper framework. Plus, whether Harris’s § 922(g)(3) con-
viction is constitutional turns on many facts unanswered by the
existing record. Although Harris admitted smoking multiple
times a week throughout the year on average, many other de-
tails are foggy. But we are not a fact-finding court. Anderson
v. Bessemer City, 470 U.S. 564, 574 (1985). And we “are a
court of review, not first view.” Frank v. Gaos, 586 U.S. 485,
493 (2019) (internal quotation marks omitted). So we will re-
mand for the District Court to fill in the record before applying
the law outlined here.
On remand, the parties should have a chance to present their
own evidence and arguments about how Harris’s drug use
affected his mental state and riskiness. In particular, the Dis-
trict Court should consider, among other factors:
• The length and recency of the defendant’s use dur-
ing and shortly before his gun possession;
• The drug’s half-life;
• Whether use of the drug affects a person’s judgment,
decision-making, attention, inhibition, or impulse
control;

-- 21 of 48 --

22
• Whether the drug may induce psychosis;
• The drug’s interference with a user’s perception of
his own impairment; and
• The long-term physical and mental effects of the use
of that drug.
We include this non-exhaustive list of factors to guide the
District Court’s inquiry into the individual defendant’s use, not
to dictate it. On remand, the District Court should explore any
questions that it thinks bear on the inquiry here. And future
courts considering § 922(g)(3) challenges should also consider
these factors in determining whether someone’s drug use sug-
gests that he “likely poses an increased risk of physical danger
to others if armed.” Pitsilides, 128 F.4th at 212 (internal quo-
tation marks omitted).
We stress that, consistent with the history, district courts
need make only probabilistic judgments of danger. A drug user
need not have harmed someone, threatened harm, or otherwise
acted dangerously to justify disarmament. But consistent with
the history, district courts must make individualized judgments
and conclude that disarming a drug user is needed to address a
risk that he would pose a physical danger to others. And in doing
so, they should consider the questions we have suggested here
and any others needed to discern whether a particular drug user
may be disarmed to prevent a risk of danger.
IV. SECTION 922(g)(3) IS NOT VAGUE
Harris also challenges the statute as vague on its face. He
objects that it does not define the phrase “unlawful user” and
so does not give “ordinary people fair notice of the conduct it

-- 22 of 48 --

23
punishes.” 18 U.S.C. § 922(g)(3); Johnson v. United States,
576 U.S. 591, 595 (2015). But that argument fails.
A. As a rule, defendants may not facially challenge
criminal laws
To challenge § 922(g)(3) as facially vague, Harris must first
show that it is vague as applied to his own conduct. The Supreme
Court has long held that, with few exceptions, a defendant
whose conduct is “clearly proscribed cannot complain of the
vagueness of the law as applied to the conduct of others.”
Holder v. Humanitarian L. Project, 561 U.S. 1, 19 (2010)
(quoting Vill. of Hoffman Ests. v. Flipside, Hoffman Ests., 455
U.S. 489, 495 (1982)). That rule “makes perfectly good sense.”
United States v. Morales-Lopez, 92 F.4th 936, 941 (10th Cir.
2024). If a statute clearly warns an ordinary person that his own
conduct is a crime, he cannot dodge liability just because it
might not be clear as to someone else.
Still, Harris claims that the Court jettisoned this bedrock
rule in Johnson. There, it invalidated the Armed Career Crim-
inal Act’s residual clause as unconstitutionally vague. 576 U.S.
at 597. It did so even though the clause was not vague “in all
its applications” and did not consider whether it was vague as
applied to the defendant’s own conduct. Id. at 603. But it did
that because of the unique problems with applying the categor-
ical approach. Morales-Lopez, 92 F.4th at 942–43. Under that
approach, courts had to ignore the defendant’s “real-world
facts” and instead hypothesize the “idealized ordinary case” of
the crime. Johnson, 576 U.S. at 597. Those intellectual gym-
nastics denied defendants fair notice about what real-world
conduct the residual clause punished. Id. at 604. But that

-- 23 of 48 --

24
provision is worlds apart from ordinary criminal laws, like
§ 922(g)(3), which depend not on hypotheticals, but on each
case’s facts.
We thus join our sister circuits in holding that Johnson did
not abrogate the ordinary rule for facial-vagueness challenges.
See United States v. Requena, 980 F.3d 30, 40–43 (2d Cir.
2020); United States v. Hasson, 26 F.4th 610, 620–21 (4th Cir.
2022); United States v. Cook, 970 F.3d 866, 877 (7th Cir.
2020); United States v. Bramer, 832 F.3d 908, 909 (8th Cir.
2016) (per curiam); Morales-Lopez, 92 F.4th at 942–43; Bowl-
ing v. McDonough, 38 F.4th 1051, 1061–62 (Fed. Cir. 2022).
Harris cannot bring his facial-vagueness attack without first
showing that the phrase “unlawful user” is vague as applied to
his case.
B. Section 922(g)(3) is not vague as applied to Harris
Harris cannot clear that bar. Though there will be border-
line cases, Harris’s habitual marijuana smoking falls squarely
within § 922(g)(3)’s plain text. So the statute put him on notice
that his conduct was a crime.
Of course, the exact boundaries of “unlawful user” are debat-
able. See Augustin, 376 F.3d at 138–39. But when a statute “can
be made constitutionally definite by a reasonable construc-
tion,” we must give it that construction. United States v. Har-
riss, 347 U.S. 612, 618 (1954). Heeding that guidance, we have
held that § 922(g)(3) requires a defendant to “have engaged in
regular use [of drugs] over a period of time proximate to or
contemporaneous with the possession of the firearm.” Augus-
tin, 376 F.3d at 139. That tracks the statute’s text. A “user” is
a “person who takes narcotic, etc., drugs,” implying some

-- 24 of 48 --

25
regularity of use. User (def. 1b), Oxford English Dictionary
(2d ed. 1989). And the verb “is” in the statute speaks in the
present tense, requiring the use of drugs to be close in time to
the gun possession.
Harris’s conduct fits the bill. He smoked unlawfully: Fed-
eral law makes marijuana illegal. 21 U.S.C. § § 812, 841. And
he did so often enough to be a “user.” He admitted smoking
marijuana at least several times per week around when he
bought the guns, including the night before buying his third
gun. So under both the statute’s “text” and “settled interpreta-
tions,” he had clear notice that he was breaking the law. United
States v. Lanier, 520 U.S. 259, 267 (1997); see also United
States v. Deng, 104 F.4th 1052, 1055 (8th Cir. 2024) (rejecting
vagueness claim by frequent marijuana smoker). His vague-
ness challenge fails.
Future cases will require closer calls about how often and
how recently one must use a drug to count as an “unlawful
user.” For instance, does a person who smokes marijuana spo-
radically before bed for chronic back pain do it regularly
enough to count? These cases will present close issues not only
of statutory construction, but also of Second Amendment
rights. But here, the question is not close. As applied,
§ 922(g)(3) is not unconstitutionally vague.
V. HARRIS’S § 922(a)(6) CONVICTIONS MUST STAND
Finally, Harris challenges his convictions for falsely deny-
ing that he was an unlawful user. At first, he claimed that there
was insufficient evidence to sustain the convictions. But his
reply brief disavowed that claim. Now, he claims only that if
we invalidate his § 922(g)(3) convictions under either the Due

-- 25 of 48 --

26
Process Clause or Second Amendment, we “must necessarily
vacate [his] § 922(a)(6) convictions” too. Reply Br. 29.
Harris’s convictions do not violate due process, so this
claim would seem to rise and fall with how the District Court
decides the Second Amendment issue on remand. But we do
not reach this claim because Harris raised it for the first time in
his reply brief. Barna v. Bd. of Sch. Directors of Panther Valley
Sch. Dist., 877 F.3d 136, 146 (3d Cir. 2017). Harris insists that
he has made the claim all along but simply mislabeled it in his
opening brief. See United States v. Negroni, 638 F.3d 434, 445
(3d Cir. 2011). Yet nothing in that brief suggests that he was
claiming, as he does in reply, that vacating his §922(a)(6) con-
victions would be “part of the remedy required” if his
§922(g)(3) convictions were found to violate the Second
Amendment. Reply Br. 29. Harris simply changed claims mid-
stream, so he did not preserve this one for appeal. His
§ 922(a)(6) convictions must stand.
*****
Common sense tells us that some mind-altering substances
make people too dangerous to trust with guns. So does our na-
tion’s regulatory tradition, which has long embraced similar
common-sense restrictions for drunks and the dangerously
mentally ill. Temporarily disarming a frequent marijuana user
like Harris may fall within that tradition. But we lack enough
facts to decide if that is so. Even so, the statutory phrase “un-
lawful user” is not vague as applied to Harris’s own frequent
marijuana use while he possessed guns. Plus, Harris’s convic-
tions for lying to get the guns must stand. So we will affirm in
part, vacate in part, and remand for the District Court to find

-- 26 of 48 --

27
facts necessary to resolve Harris’s Second Amendment
challenge.

-- 27 of 48 --

1
KRAUSE, Circuit Judge, concurring, with whom BIBAS,
Circuit Judge, joins.
I join the majority opinion in full. As the majority
persuasively explains, while habitual marijuana use was
virtually nonexistent at the Founding, early legislatures
wrestled with analogues concerns, namely drunkenness and
lunacy. See Maj. Op. 7–14. And they imposed similar, if not
more burdensome, restrictions on drunks and lunatics than
those which 18 U.S.C. § 922(g)(3) imposes on users of
controlled substances. I write separately, however, with some
observations about our Nation’s evolving—and conflicted—
relationship with marijuana and how modern-day
understandings may inform the application of § 922(g)(3) to
habitual marijuana users.
Our regulations concerning marijuana have shifted over
time with our developing uses of it and our understanding of
its properties. As the majority recounts, early uses of hemp
were limited to cloth, paper, and rope. See id. at 6. It was not
until the early twentieth century that people began smoking
marijuana recreationally. Martin Booth, Cannabis: A History
127–28 (2003). Early regulation—coming off the heels of the
temperance movement—primarily consisted of state and local
enforcement, until the federal government in 1937 stepped in
with the Marihuana Tax Act, which imposed “onerous
administrative requirements” and “prohibitively expensive
taxes” that “practically curtailed the marijuana trade.”
Gonzales v. Raich, 545 U.S. 1, 11 (2005).
By the mid- to late-1960s, however, public sentiment
had changed, with recreational marijuana use budding among
young people. See James B. Slaughter, Marijuana Prohibition

-- 28 of 48 --

2
in the United States: History and Analysis of a Failed Policy,
21 Colum. J.L. & Soc. Probs. 417, 420 (1988). And by the
mid-1970s, marijuana gained wider acceptance as a
recreational drug. Booth at 240. These changes in Americans’
attitudes towards marijuana prompted states to revise their
regulations, so that by 1973, nearly every state had
substantially reduced its penalties for simple marijuana
possession. Richard J. Bonnie & Charles H. Whitebread, The
Marijuana Conviction: A History of Marijuana Prohibition in
the United States 240, 278–79 (1999). Much of this
deregulation has been chalked up to the congressionally
directed National Commission on Marihuana and Drug
Abuse’s report, Marihuana: A Signal of Misunderstanding.
See Slaughter at 422–24. Its comprehensive investigation into
marijuana concluded, among other things, that “there is little
proven danger of physical or psychological harm from the
experimental or intermittent use of the natural preparations of
cannabis,” and that “its use at the present level does not
constitute a major threat to public health.” Nat’l Comm’n on
Marihuana and Drug Abuse, Marihuana: A Signal of
Misunderstanding 65, 90 (1972).
The federal government, however, remained
unpersuaded. Recognizing its psychotropic effects, Congress
prohibited the manufacture, distribution, dispensation, and
possession of marijuana in 1970, replacing the preceding
patchwork of various state and federal regulation.1 21 U.S.C.
§§ 841(a)(1), 844(a). This federal prohibition presaged a
1 While the federal government did regulate marijuana in
various forms before 1970, the Controlled Substances Act was
the first comprehensive federal law targeting illicit drug use and
possession. See Gonzales v. Raich, 545 U.S. 1, 10–14 (2005).

-- 29 of 48 --

3
retrenchment in public perceptions of marijuana in the 1980s,
triggered by “[a]dolescent marijuana use, the appearance of
cocaine use and abuse on a national scale[,] and the rising
potency of marijuana.” Slaughter at 438–39. This pendulum
swing catalyzed renewed federal enforcement of marijuana
prohibitions and additional federal criminal legislation
targeting the drug trade. See id. at 443–46; Comprehensive
Crime Control Act of 1984, Pub. L. 98-473, 98 Stat. 1976.
Today, marijuana is legal to various extents in forty
states, including for recreational use in twenty-four states and
the District of Columbia. State Medical Cannabis Laws, Nat’l
Conf. of State Legislatures (June 27, 2025),
https://www.ncsl.org/health/state-medical-cannabis-laws.
Federal law, however, has continued to bar its use, listing it in
the Controlled Substances Act as a Schedule I controlled
substance, 21 U.S.C. § 812, sched. I(c)(10), meaning that in the
view of Congress and the Attorney General, id. § 811(a), it
“has a high potential for abuse,” “has no currently accepted
medical use in treatment in the United States,” and lacks
“accepted safety for use of the drug or other substance under
medical supervision,” id. § 812(b)(1). And because marijuana
is classified as a controlled substance, its users are subject to
§ 922(g)(3)’s ban on possession of a firearm by habitual users.
18 U.S.C. § 922(g)(3).
That brings us to the case presently before us, for even
as some have pressed Congress to follow the states’ lead,
modern science has prompted a reassessment of that more
permissive approach to marijuana use, see, e.g., Rosalie
Liccardo Pacula et al., Developing Public Health Regulations
for Marijuana: Lessons from Alcohol and Tobacco, 104 Am.
J. Pub. Health 1021, 1022 (2014) (recognizing the growing

-- 30 of 48 --

4
consensus about “certain acute effects and consequences of
chronic [marijuana] use” warranting public health regulation),
with implications for as-applied challenges by marijuana users.
Clinical studies reflect that marijuana use impairs users’
judgment, decision-making, attention, and inhibition, Michael
L. Alosco et al., Neuropsychology of Illicit Drug Use and
Impulse Control Disorders, in Clinical Neuropsychology: A
Pocket Handbook for Assessment 605, 608 (Michael W.
Parsons & Thomas A. Hammeke eds., 3d ed. 2014),2 causing
symptoms that mirror those of mild cognitive impairment that
might arise from mental illness or alcohol, see Cognitive
Impairment, Nat’l Cancer Inst. Dictionary,
https://www.cancer.gov/publications/dictionaries/cancer-
terms/def/cognitive-impairment (last visited July 10, 2025).
According to some studies, these effects can last for hours after
use, Alosco et al. at 608, and in frequent cannabis users, they
can be more intense and last longer, Rebecca D. Crean et al.,
An Evidence Based Review of Acute and Long-Term Effects of
Cannabis Use on Executive Cognitive Functions, 5 J.
Addiction Med. 1, 5–6 (2011).
2 See also Harmelin v. Michigan, 501 U.S. 957, 1002 (1991)
(Kennedy, J., concurring) (observing the danger posed by
“drug-induced changes in physiological functions, cognitive
ability, and mood”); Nat’l Acads. of Scis., Eng’g, and Med.,
The Health Effects of Cannabis and Cannabinoids: The
Current State of Evidence and Recommendations for Research
53 (2017) (explaining that the effects of marijuana intoxication
include an altered “perception of time,” “decreased short-term
memory,” and “impaired perception and motor skills,” and
that, at higher doses, marijuana can cause “panic attacks,
paranoid thoughts, and hallucinations”).

-- 31 of 48 --

5
Other studies reflect that this impaired judgment,
diminished motor skills, and lower inhibition make it
dangerous to combine marijuana with high-risk activities. See,
e.g., Thomas D. Marcotte et al., Driving Performance and
Cannabis Users’ Perception of Safety, 79 JAMA Psychiatry
201, 206 (2022) (finding that even though people may not
perceive that they are unsafe drivers an hour and a half after
smoking marijuana, they perform worse on driving
simulators); Daniel T. Myran et al., Cannabis-Involved Traffic
Injury Emergency Department Visits After Cannabis
Legalization and Commercialization JAMA Network Open,
Sept. 6, 2023, art. e2331551, at 7,
https://jamanetwork.com/journals/jamanetworkopen/fullarticl
e/2808961 (finding that, between 2010 and 2021, cannabis-
involved traffic accidents in Ontario requiring emergency-
room treatment rose 475%).
Some studies show that frequent marijuana use can
prolong these consequences because smoking marijuana
chronically causes THC, which gives marijuana its
psychoactive properties, to build up in the blood, potentially
contributing to longer-lasting cognitive effects. Emese Kroon
et al., Heavy Cannabis Use, Dependence, and the Brain: A
Clinical Perspective, 115 Addiction 559, 565 (2019). One
meta-analysis found that chronic marijuana use can impair
decision-making, increase risk-taking, and exacerbate
impulsivity for hours, days, or even a few weeks. See Crean et
al. at 3 tbl. 2, 4–5. (“[C]hronic, heavy cannabis use[rs]
show . . . enduring deficits following three weeks or more
abstinence” in decision-making and other executive
functions.).

-- 32 of 48 --

6
Notably, the marijuana currently available for
consumption may magnify these risks, as the marijuana
available today is far more potent than it was several decades
ago, containing about four times as much THC. Compare
Mahmoud A. ElSohly et al., Changes in Cannabis Potency
over the Last Two Decades (1995–2014): Analysis of Current
Data in the United States, 79 Biological Psychiatry 613, 613
(2016) (reporting 4% THC concentration in 1995), with Suman
Chandra et al., New Trends in Cannabis Potency in USA and
Europe During the Last Decade (2008–2017), 269 Eur.
Archives Psychiatry & Clinical Neuroscience 5, 9 (2019)
(reporting increase to 17% THC concentration by 2017).
I agree with the majority that § 922(g)(3), as applied to
“those whose drug use would likely cause them to pose a
physical danger to others if armed,” Maj. Op. 6, is “consistent
with this Nation’s historical tradition” of regulating gun
possession by drunks and lunatics, N.Y. State Rifle & Pistol
Ass’n v. Bruen, 597 U.S. 1, 17 (2022), and reflects the type of
common-sense prophylactic judgment that the Second
Amendment permits, see Maj. Op. 14–15. We leave it to the
District Court on remand to develop the record on which it
relies to apply the statute to Harris. Given the cognitive and
motor impairments associated with marijuana use, however,
our evolving understanding of the effects of marijuana, as
reflected in these types of scientific studies, will bear heavily
on that determination.

-- 33 of 48 --

1
AMBRO, Circuit Judge, concurring in part and dissenting in
part
Erik Harris has no history of violence or threatening
behavior. The Government wants to disarm him anyway under
18 U.S.C. § 922(g)(3) because he used marijuana around the
time he bought his gun. Harris argues that charging him under
that provision violates his Second Amendment right. I join my
colleagues on a remand that would require the District Court to
determine whether the Government has proven that Harris’s
marijuana causes him to “pose[] a clear threat of physical
violence to another” before the Government can disarm him.
Pitsilides v. Barr, 128 F.4th 203, 209 (3d Cir. 2025) (Krause, J.)
(quoting United States v. Rahimi, 602 U.S. 680, 698 (2024)).1
My colleagues correctly note that Harris may be disarmed if
his marijuana use makes him a “credible threat to the physical
safety of others with a gun.” Maj. Op. 14 (citing Rahimi, 602
U.S. at 694, 698). But they also obscure this conclusion with
language that sets the threshold for potential dangerousness too
low. See, e.g., id. at 16 (“likely pose[] an increased risk of
physical danger to others if armed” (internal quotation marks
omitted)), 19 (alluding to an undefined “likelihood” of risk),
20 (suggesting the District Court need only examine whether
“Harris’s frequent marijuana use increased the risk that he
could not handle guns safely”). No other court does so, and on
this I dissent.
1 I also join the portion of the decision that affirms Harris’s
convictions under 18 U.S.C. § 922(a)(6) and the District
Court’s denial of his constitutional vagueness challenge to
§ 922(g)(3).

-- 34 of 48 --

2
I
Any modern statutory ban on firearms must be
“consistent with the principles that underpin the Nation’s
regulatory tradition.” Rahimi, 602 U.S. at 681 (citing N.Y. State
Rifle & Pistol Ass’n v. Bruen, 597 U.S. 1, 26–31 (2022)). A
two-step framework guides our analysis.
At the first step, we ask whether the Second
Amendment presumptively protects Harris’s conduct. All
agree that it does. Harris is among “the people” who have a
right “to keep and bear [a]rms.” U.S. Const. amend. II. And he
was convicted under § 922(g)(3) for “quintessential Second
Amendment conduct: possessing a handgun.” Maj. Op. 5
(quoting United States v. Moore, 111 F.4th 266, 269 (3d Cir.
2024), cert. denied, 2025 WL 1787742 (2025)).
At the second step, the Government must prove that
disarming Harris is “consistent with the Nation’s historical
tradition of firearm regulation.” Range v. Att’y Gen., 124 F.4th
218, 228 (3d Cir. 2024) (en banc) (quoting Bruen, 597 U.S. at
24). To carry its burden, the Government must show that
§ 922(g)(3) is “relevantly similar” to historical “analogue[s],”
which are Founding-era laws that “impose[d] a comparable
burden” with a “comparabl[e] justifi[cation].” Bruen, 597 U.S.
at 29–30. “[I]f laws at the [F]ounding regulated firearm use to
address particular problems, that will be a strong indicator that
contemporary laws imposing similar restrictions for similar
reasons” are permissible. Rahimi, 602 U.S. at 692. But “[e]ven
when a law regulates arms-bearing for a permissible
reason, … it may not be compatible with the right if it does so
to an extent beyond what was done at the [F]ounding.” Id.
Case law has distilled three additional rules. First, and
most important, “the Second Amendment’s touchstone is
dangerousness.” Pitsilides, 128 F.4th at 210 (Krause, J.)

-- 35 of 48 --

3
(quoting Folajtar v. Att’y Gen., 980 F.3d 897, 924 (3d Cir.
2020) (Bibas, J., dissenting)). The consistent theme uniting
Founding-era restrictions is that individuals who pose a threat
to the physical safety of others can be disarmed. Rahimi, 602
U.S. at 693, 697 (addressing a challenge to § 922(g)(8)); see
also Range, 124 F.4th at 230 (explaining that Range could not
lose his Second Amendment rights without “evidence that he
pose[d] a physical danger to others”). Other circuits to have
addressed the question here agree. See United States v. Cooper,
127 F.4th 1092, 1095 (8th Cir. 2025) (“[F]or disarmament of
drug users and addicts to be comparably justified [with
Founding-era laws], it must be limited to those who pose a
danger to others.” (brackets and internal quotation marks
omitted)); United States v. Connelly, 117 F.4th 269, 277 (5th
Cir. 2024) (requiring a factfinder to determine whether the
defendant “presents a danger to [himself] and others”).
Second, we know that “[n]either our historical tradition
nor our modern understanding of the Second
Amendment … permits us to blindly defer to a categorical
presumption that a given individual permanently presents a
special risk of danger.” Range, 124 F.4th at 276 (Krause, J.,
concurring in the judgment); see also id. at 230 (majority
rejecting a “categorical argument” that all “those convicted of
serious crimes” could be “expected to misuse firearms”
because it was “far too broad” (internal quotation marks
omitted)). At the Founding, the state could burden a person’s
Second Amendment right only after an individualized
determination that he posed a physical danger to others. So
today’s as-applied dangerousness inquiry is common sense: if
the standard for disarming drug users is dangerousness, and
countless marijuana users are not dangerous, then not every
marijuana user can be stripped of his gun rights. Categories are
out. Individual assessments are in.

-- 36 of 48 --

4
Finally, we know that bans on gun possession by drug
users must be temporary. A person may cease to be dangerous.
See Rahimi, 602 U.S. at 698–99. When he is no longer
dangerous, he gets his rights—and guns—back. See id. at 699.
And, per the majority, “[s]omeone who regularly uses mind-
altering substances that make him a ‘credible threat to the
physical safety of others with a gun’ may be disarmed
temporarily until he stops using drugs.” Maj. Op. 14 (quoting
Rahimi, 602 U.S. at 694, 698).
II
So what relevantly similar principle at the Founding
justifies applying § 922(g)(3) to Erik Harris? My colleagues
agree that it has no historical twin. Maj. Op. 6. Instead, they
cite Founding-era “regulations on the dangerously drunk and
dangerously mentally ill.” Id. But Founding-era lawmakers did
not have free rein to disarm people who drank alcohol. The key
to this “permissible reason,” Rahimi, 602 U.S. at 692, was
dangerousness caused by intoxication. See Maj. Op. 7
(explaining the Founding generation “understood that drinking
could provoke people to act dangerously”).
For example, regulations on the “dangerously drunk”
punished people who were “‘abus[iv]e to’ or ‘striking’ others.”
Id. at 8 (quoting General Laws and Liberties of the
Massachusetts Colony 81 (1672)). They also restricted the gun-
rights of those who “abused” or “revil[ed]” officers, id. at 9
(first citing Laws of the State of Delaware (New Castle, Del.,
Samuel and John Adams 1797); and then quoting A Digest of
the Laws of Maryland (Baltimore, Thomas Herty ed., 1799)),
or who “terrif[ied] or disquiet[ed] the good People of th[e]
State,” An Act Against Breaking the Peace, reprinted in Acts
and Laws of the State of Connecticut, in America 189
(Hartford, Elisha Babcock (1786)). The throughline

-- 37 of 48 --

5
connecting these laws is that they disarmed drunkards who
“pose[d] a danger to others.” United States v. Veasley, 98 F.4th
906, 916 (8th Cir. 2024), cert. denied, 145 S. Ct. 304 (2024).
True, not all regulations punished “abusive” drunks. But
there is no need to “cherry pick” history. Maj. Op. 17. As my
colleagues suggest, the regime as a whole was aimed to ensure
good behavior, prevent drunks from breaking the peace, and
address threats “to the physical safety of others,” id. at 14; see
id. at 15 (“Founding-era legislatures often required drunks to
post bonds … based on the judgment that drunks posed a risk
to public peace and the safety of others.”). The principle I thus
draw is that Founding-era laws targeted dangerous behavior
that followed from intoxication; so today those who pose a
“credible threat to the physical safety of others” because of
their intoxication may be disarmed. Rahimi, 602 U.S. at 693,
700.
The majority’s analogy to laws regulating “lunatics” at
the Founding is more strained. True, “[s]ociety’s answer to
mental illness … was to lock up anyone who was dangerous or
disturbing to others.” Veasley, 98 F.4th at 915 (internal
quotation marks omitted). But while “[e]arly in this country’s
history[] the ‘mentally ill and dangerous’ ended up in jails,
makeshift asylums, and mental hospitals ‘with straitjackets and
chains[,]’ … ‘[t]hose who posed no danger’ … ‘stayed at
home with their families,’ with ‘their civil liberties … intact.’”
Cooper, 127 F.4th at 1095 (quoting Veasley, 98 F.4th at 913,
915).
Neither the majority nor the Government credibly
explains how marijuana users resemble the dangerously
mentally ill. Without a much stronger connection between
marijuana use and dangerousness of the kind posed by those
with serious mental illness, we cannot use the rationale

-- 38 of 48 --

6
underlying Founding-era laws regulating those individuals to
justify § 922(g)(3)’s regulation of marijuana users. See
Connelly, 117 F.4th at 276 (“The government highlights
nothing demonstrating that laws designed to confine (and
consequently, disarm) those so severely mentally ill that they
presented a danger to themselves and others map onto
§ 922(g)(3)’s rationale.”); United States v. Harrison, 654 F.
Supp. 3d 1191, 1214 (W.D. Okla. 2023) (“There are likely
nearly 400,000 Oklahomans who use marijuana under state-
law authorization. Lumping all those persons into a category
with ‘dangerous lunatics[]’ … is a bridge too far.”).
With this background, and assuming an analogy
between alcohol intoxication and frequent marijuana use works
when a marijuana user poses a threat of physical violence to
another while armed, I abide a remand. No doubt
dangerousness is the touchstone. Contra Maj. Op. 17
(suggesting I disagree with a “test focused on risk of danger”).
Whether we call it a “clear threat,” “credible threat,” or
something else, the Government must show by a
preponderance of the evidence that Harris’s drug use and
weapon possession make him a physical danger to others.
Though my colleagues dispute my historical overview,
they do not clearly explain if they draw a different principle or,
if so, what it is. They instead give us varying dangerousness
thresholds for one who is armed: id. at 3 (“pose a special
danger of misusing firearms”); id. at 14, 16, 18 (credible
threat); id. at 19 (likely poses a danger); id. at 16, 20 (“likely
poses an increased risk of physical danger”); id. at 20
(“increased … risk”). It is the “increased risk” formulation that
is both off-point and concerning. To the majority, someone who
uses marijuana can be disarmed even if he plainly is not
dangerous, so long as his use increases the chance he could act

-- 39 of 48 --

7
dangerously. No Founding-era analogue justifies restricting
gun possession in that circumstance.
That lower threshold of increased risk ignores the bar
the Supreme Court and our Court have established. The former
in Rahimi spoke of a physical threat being “clear,” 602 U.S. at
698, “credible,” id. at 698-702, or “demonstrated,” id. at 698,
before the Government could disarm a person.
As for our Court, the en banc decision in Range
borrowed from Rahimi that threats to others had to be “clear”
or “credible.” Range, 124 F.4th at 230. Or consider Pitsilides,
a recent application of the dangerousness standard. We
explained that Rahimi and Range showed “disarmament is
justified as long as a felon … ‘present[s] a special danger of
misusing firearms,’ in other words, when he would likely ‘pose
a physical danger to others’ if armed.” Pitsilides, 128 F.4th
at 210 (first quoting Rahimi, 602 U.S. at 698; and then quoting
Range, 124 F.4th at 232) (brackets omitted). Pitsilides had
operated an illegal gambling ring on and off for more than a
decade. Id. at 205–06. He staffed his games with security, and
SWAT teams had broken them up. Id. at 206. Plus he had a
criminal record related to his gambling offenses. Id. at 206,
212. Adding to that, we had law establishing a link between
gambling and organized crime. Id. at 213 (citing United States
v. Williams, 124 F.3d 411, 417 n.7 (3d Cir. 1997) (noting
“Congress[’]s recognition that gambling has historically
provided a major source of revenue for organized crime
groups”)). Even on that record, we could not answer whether
Pitsilides, if armed, met the dangerousness standard from
Rahimi and Range, so we remanded for the District Court to
determine whether he “pose[d] a special danger of misusing
firearms in a way that would endanger others.” Id. at 213.

-- 40 of 48 --

8
In making that determination, we deemed “crucial” a
court’s “consideration of” a defendant’s “post-conviction
conduct” that might “indicat[e] … dangerousness.” Id. at 212.
That is because “such conduct may be highly probative of
whether an individual likely poses an increased risk of physical
danger to others if armed.” Id. But that was not our holding.
We ended our opinion by emphasizing that we remanded to fill
“gaps in [the] record” with “additional discovery of facts
probative to the prevailing Second Amendment analysis,
including whether Pitsilides poses a special danger of misusing
firearms in a way that would endanger others.” Id. at 213
(emphasis added). We never suggested that the Government
could disarm Pitsilides if his gambling activity or criminal
record merely “increased the risk that he could not handle guns
safely.” Maj. Op. 20. We did not, nor could we, expand the
holdings of Rahimi and Range in that way.
Moreover, the majority says nothing about the level to
which a person’s risk must rise before the Government can
disarm him. Could a person whose risk increases from
negligible to slightly more than negligible be disarmed under
this test? Theoretically, most people pose a slightly greater risk
of danger with a gun while intoxicated than while sober. So
what the majority calls an individualized dangerousness
inquiry begins to look like a categorical rule in disguise. As
explained above, however, the Second Amendment rarely
tolerates categorical rules. Range, 124 F.4th at 276 (“Neither
our historical tradition nor our modern understanding of the
Second Amendment … permits us to blindly defer to a
categorical presumption that a given individual permanently
presents a special risk of danger.”) (Krause, J., concurring in
the judgment).

-- 41 of 48 --

9
My colleagues respond by arguing that Rahimi, Range,
and Pitsilides do not control the outcome here because they
invoke different historical traditions—those justifying
§§ 922(g)(1) (regulating gun possession of felons) and (g)(8)
(regulating gun possession of those subject to restraining
orders). Maj. Op. 18–19. But then they reach the conclusion I
espouse. They concede their cited history “operated like the
surety and going-armed laws that the Supreme Court blessed
in Rahimi.” Id. at 16. And they purport to adopt—though not
consistently—principles articulated in those cases. E.g., id. at 3
(adopting the “pose a special danger of misusing firearms”
principle from Pitsilides); 14 (adopting the “credible threat”
standard in Rahimi and Range). How do those standards fit
holding here a lower standard that requires only an increased
risk of danger?
The also say that the relevant tradition here—disarming
the dangerously intoxicated and the dangerously mentally ill—
sanctioned predictive judgments about dangerousness even
before someone got hurt. Constables, after all, did not need to
wait until an intoxicated person injured somebody to disarm
him. I agree. But there is a difference between (a) disarming
someone who presents a clear threat of danger to others based
on his behavior before he has harmed another person and (b)
disarming someone because he poses some undefined level of
risk. The tapestry of historical regulation yields a clear
principle: people could be disarmed at the Founding when they
posed a danger to others because of their intoxication. By
holding that the Government can disarm someone even when
he does not pose a clear threat of physical violence to another
by a preponderance of the evidence, the majority draws a
principle unsupported by history and tradition.

-- 42 of 48 --

10
III
My colleagues write separately to expound their views
on the dangers posed by marijuana. But that separate writing
contains virtually no legal reasoning and gives almost no sense
that it is meant to. It instead reads like a policy statement to
Congress advocating for a marijuana ban. Using select non-
record sources, my colleagues draw their own conclusions
about marijuana’s effects on users’ judgment, decision-making,
attention, and inhibition, the duration of its alleged effects on
cognition, and its potency.
For instance, they cite a meta-analysis for the
proposition that “chronic marijuana use can impair decision-
making, increase risk-taking, and exacerbate impulsivity for
hours, days, or even a few weeks.” Conc. 5 (citing Rebecca D.
Crean et al., An Evidence Based Review of Acute and Long-
Term Effects of Cannabis Use on Executive Cognitive
Functions, 5 J. Addiction Med. 1, 3 tbl. 2, 4–5 (2011)). They
speculate that “the marijuana currently available for
consumption may magnify these risks, as the marijuana
available today is far more potent than it was several decades
ago.” Id. at 6 (citing Suman Chandra et al., New Trends in
Cannabis Potency in USA and Europe During the Last Decade
(2008–2017), 269 Eur. Archives Psychiatry & Clinical
Neuroscience 5, 9 (2019)). Symptoms from marijuana use,
they claim, mirror those of “mild cognitive impairment that
might arise from mental illness or alcohol.” Id. at 4 (Their
support for this claim is a dictionary definition for “cognitive
impairment” from the National Cancer Institute that does not
mention marijuana. Id.)
We are judges—not scientists, sociologists, or
policymakers. Parsing scientific evidence in the first instance
is not our role, and we generally are not good at it. This kind

-- 43 of 48 --

11
of freewheeling appellate factfinding is inappropriate, see
Amadeo v. Zant, 486 U.S. 214, 228 (1988) (reversing and
remanding when “the Court of Appeals … engage[d] in
impermissible appellate factfinding”), in part because it
relieves the Government of its duty to “affirmatively prove[]
that [§ 922(g)(3)] is ‘consistent with the Second Amendment’s
text and historical understanding,’” Rahimi, 602 U.S. at 737
(Barrett, J., concurring) (quoting Bruen, 597 U.S. at 26).
As my colleagues note, marijuana is legal in some form
in 39 states and the District of Columbia. For better or worse,
our Nation’s democratic policymaking process has gradually
liberalized laws regulating marijuana over the past few
decades. I take no position on the wisdom of this trend because
I am a judge, not a legislator. My colleagues have deeply-held
and good-faith views about marijuana, but those views are the
stuff of policy, not law, and they would be better aired in an op-
ed than in the Federal Reporter.
IV
Subsection 922(g)(3) is constitutional as applied to
Harris only if his marijuana use makes him a “clear threat of
physical violence to another.” Pitsilides, 128 F.4th at 209
(quoting Rahimi, 602 U.S. at 698). I go no further.2 “Future
cases may present other and more difficult questions …. But
we take cases as they come and today [should] resolve only the
question posed to us.” Bondi v. VanDerStok, 145 S. Ct. 857,
2 The majority identifies several factors for the District Court to
consider on remand, like marijuana’s “half-life” and
“[w]hether the drug may induce psychosis.” Maj. Op. 21.
These may be relevant to the ultimate dangerousness inquiry,
but almost none of the considerations they outline sheds light
on whether Harris himself was dangerous.

-- 44 of 48 --

12
869–70 (2025). It is a “fundamental principle of judicial
restraint … that courts should neither anticipate a question of
constitutional law in advance of the necessity of deciding it nor
formulate a rule of constitutional law broader than is required
by the precise facts to which it is to be applied.” Wash. State
Grange v. Wash. State Republican Party, 552 U.S. 442, 450
(2008) (internal quotation marks omitted).
As for Harris, here is what we know. He was a college
student in 2019 with no history of violence and no prior arrests.
He bought a total of three guns in February and March of that
year. With each purchase he answered “no” on a federal form
asking whether he was a user of or addicted to marijuana. He
was, however, a regular user at that time.
Five days after he bought the second gun, Harris and his
childhood friend, Jaemere Scott, celebrated Scott’s mother’s
birthday at Scott’s home and later at a bar. Harris drank alcohol
and smoked marijuana that evening. When the two arrived at
the bar, Scott asked Harris whether he had his gun on him,
warning him not to bring it into the bar. Harris did not, thinking
he had left it in his car, and they entered the bar. But when
Harris and Scott left, Harris realized that the gun was not in his
car after all. He went to his girlfriend’s house, checked there
for the gun, could not find it, and went to sleep. The next day,
he searched Scott’s house, another friend’s house, and his car
once more, but the gun was still missing. Harris then returned
to the bar to look for it. Coming up empty, he called the police
and reported the gun stolen. The police ultimately found it with
Scott, a convicted felon. Suspecting Harris had purchased the
gun for Scott, they interviewed Harris. During the interview,
he denied doing so but admitted being a frequent user of
marijuana.

-- 45 of 48 --

13
The Government, joined by my colleagues in the
majority, would disarm Harris because he could not locate his
gun after smoking marijuana. But that gun was found with
someone Harris grew up with and was close to personally.
From the record, we do not know when (or if) he misplaced it,
whether he was high at that time, or how it ended up with Scott
(who may have stolen it). Despite its early suspicions that
Harris had purchased the gun for Scott, the Government did not
indict Harris for that conduct. In my view, nothing in the record
before us suggests that he poses a danger to the physical safety
of others.
V
Why not stick to the Supreme Court’s Second
Amendment decisions, our en banc decision in Range, or this
very panel’s holding in Pitsilides? An unlawful drug user may
be disarmed if he poses a credible threat to the physical safety
of others with a gun—that is, if it is more likely than not that
his drug use paired with gun possession makes him dangerous.
The waters are roiled enough that we need a breather (awaiting
further clarity from the Supreme Court) to sort things out.
Instead, we get yet another test—what matters this time is not
dangerousness but any whiff of its increased risk, suggesting a
lower threshold than before.
Many, if not most, readers of this partial dissent know
someone who uses marijuana—maybe a sick friend who uses
it to treat pain, an insomniac relative who uses it to sleep at
night, a veteran who uses it to manage his post-traumatic stress
disorder, or hunters in a duck blind.3 Were intoxication
3 My colleagues say that a “buzzed brain with a loaded gun
sounds like a misfire waiting to happen.” Maj. Op. 20. But their

-- 46 of 48 --

14
minimally to increase the risk of dangerousness associated
with possessing a gun, it is hard to imagine a marijuana user
whom the majority’s policy-made test would not lump together
with dangerous drunks and “lunatics.” Indeed, the majority
states categorically that “[c]ommon sense tells us that
[marijuana] make[s] people too dangerous to trust with guns.”
Maj. Op. 26. The consequence of that reasoning could be that
most of these individuals, along with countless American
adults, are vulnerable to disarmament. That should give us
pause. If our reasoning authorizes legislatures to suspend the
constitutional rights of so many for such common behavior, it
may mean that we are not taking the Supreme Court’s
instruction seriously and are instead drawing a “principle at
such a high level of generality that it waters down the right.”
Rahimi, 602 U.S. at 740 (Barrett, J., concurring).
The majority leaves us with an amorphous holding that
flouts precedent, defies common sense, and creates a circuit
split. Gun possession and marijuana use may at times be a
“lethal cocktail,” Maj. Op. 2, but those times are scattered in a
mountain of mismatches. I concur in the judgment only to the
extent that it affirms Harris’s convictions under 18 U.S.C.
§ 922(a)(6) and the District Court’s denial of his constitutional
vagueness challenge to § 922(g)(3). Because the majority
instructs the District Court to consider Harris’s increased risk
folksy retort gives the game away. If we accept that simply
being intoxicated is enough to be disarmed, without some
individualized determination that the user would be dangerous
because of his intoxication, then we are endorsing a
disarmament regime based on categorical dangerousness
judgments. In the majority’s view, if you drink, then you can
be disarmed. That was certainly not the historical tradition at
the Founding.

-- 47 of 48 --

15
of dangerousness rather than his actual threat of danger to
others caused by his marijuana use, the tipping point is too low.
Thus I respectfully dissent.

-- 48 of 48 --