You have seen the claim. Ninety percent of guns used in crimes are obtained illegally, so gun laws only burden people who follow them. It travels with a federal citation attached, and the citation is real.
So this show went and read it. The number is in there. The word "illegally" is not.
The Receipt
The source is a January 2019 special report from the Bureau of Justice Statistics, Source and Use of Firearms Involved in Crimes: Survey of Prison Inmates, 2016, by two of the Bureau's statisticians, Mariel Alper and Lauren Glaze. Between January and October of 2016, interviewers sat face to face with a nationwide sample of people held in state and federal prisons. About one in five, 21 percent, said they had a gun during the crime that put them there. The Bureau estimates that at 287,400 people, and 256,400 of them said how they got it.
Table 6 is where the famous number lives. One row reads "Not purchased or traded at retail source," and the figure beside it is 89.9 percent. The row under it, purchased or traded at a retail source, is 10.1 percent. The report defines a retail source as a gun shop or store, a pawn shop, a flea market, or a gun show.
That is the whole finding. Nine in ten did not get the gun over a retail counter. Whether they got it lawfully is a different question, and here is what most retellings miss: the firearm questions the report prints never ask it.
The report prints its firearm questions in an appendix, and the two that built these tables are plain. "How did you obtain the firearm." "Where did you obtain" it. None of the firearm questions in that appendix asks whether the transfer was legal, or whether the person was allowed to own a gun at all. The closest it comes is a follow-up asking prisoners why somebody else bought the gun for them, with "not allowed because of criminal record" as one of the boxes.
Here is what they did say, from Table 5. Off the street or the underground market: 43.2 percent. From an individual, meaning bought, borrowed or received as a gift from family or a friend: 25.3 percent. Theft: 6.4 percent. Other sources, including a gun found at the scene or brought by somebody else: 17.4 percent. A gun show: 0.8 percent. Prisoners could name more than one source, so the rows do not add to a hundred.
The Bureau itself calls the underground market "illegal sources," meaning markets for stolen goods, fences, criminal enterprises and drug sellers. And it does its own combining on page one: "more than half (56%) had either stolen it (6%), found it at the scene of the crime (7%), or obtained it off the street or from the underground market (43%)." That is the federal number for the clearly dirty channels. Fifty-six, not ninety. For the quarter who got a gun from family or a friend, the report does not say whether the handoff broke any law. It could not say. The firearm questions it prints did not ask.
The switch
Two different questions hide inside that viral sentence, and the claim only works if you never notice the swap.
The first question is how the last person got the gun. That is what the prison survey measures, by self-report, among people who were caught, convicted and locked up. About 10 percent of state prisoners and 14 percent of federal prisoners who had a gun gave no usable answer on the source and are left out of the tables. It is also the newest there is. The Bureau's page for the Survey of Prison Inmates lists 2016 as the last collection year.
The second question is where the gun entered the market in the first place. The survey cannot answer that one. A different federal record can.
How the machine works
When police recover a gun and want its history, they ask the Bureau of Alcohol, Tobacco, Firearms and Explosives to trace it to the last known sale by a licensed business. The agency's National Firearms Commerce and Trafficking Assessment publishes the totals, and Volume IV carries the trace data through 2023.
For 2017 through 2023, ATF counted 2,294,719 traced crime guns for which it knew both the buyer and the kind of licensee that made the sale. Licensed dealers had transferred 69 percent of them. Licensed pawnbrokers, 23 percent. Licensed manufacturers, 7 percent. That is about 99 percent of those guns, the ones ATF traced to a known buyer and a known kind of licensee, walking out the front door of a federally licensed business at some point in their lives.
Then they changed hands. In 2022 and 2023, ATF reports, nearly 56 percent of traced crime guns had a purchaser who was not the person found with the gun. In another 31 percent, no possessor was identified at all. The purchaser and the possessor were the same person for, in the agency's words, "only about 13%."
So ask the two questions this show always asks. Who decided. And who got paid. Every one of those guns crossed a licensed counter once, and a licensed business rang it up. The record shows a store, then a gap, then a crime scene. The argument is over the gap.
ATF measures the gap in time. It calls the span between a gun's last known sale and its recovery "time-to-crime," and it writes that, in its investigative experience, a time-to-crime under three years "can be an indicator of illegal firearm trafficking." In 2023, 49 percent of traced crime guns came in under three years and 20 percent came in under one.
The agency also went through 9,708 of its own closed trafficking investigations from 2017 to 2021 and sorted them by channel. The two biggest were unlicensed dealing by private persons, in 40.7 percent of cases, and straw purchasing, in 39.5 percent. ATF describes the channel as "straw purchasers who buy firearms for criminal sellers and users." The categories overlap. Unlicensed dealers averaged about 20 guns per case, straw buyers a little more than 11. Licensed dealers doing the trafficking themselves figured in 1.6 percent.
And then there is plain theft, which is the big one. ATF's theft analysis counts 1,062,970 firearms reported stolen from 2019 through 2023. About 3 percent were taken from licensed dealers. Almost 2 percent vanished from interstate shipments. Ninety-five percent, 1,013,964 guns, were stolen from private citizens. That is a floor. The same report notes that federal law and most states do not require a gun owner to report a theft, and cites a Bureau of Justice Statistics survey finding that only about 75 percent of private gun thefts get reported.
Why it costs you, and why Florida is in this story
Nine of the ten states with the highest rates of guns stolen from private citizens are in the South, by ATF's count. Florida's rate is lower than Georgia's or Alabama's, and the raw number is still large: an average of 13,496 firearms reported stolen from private citizens in Florida every year from 2019 through 2023. ATF lists fifteen states that, as of 2022, required owners to report a stolen gun. Florida is not on the list.
Florida shows up again on the other end. ATF describes an "East Coast I-95 Pipeline" and names Florida among its six source states, with Georgia, the Carolinas, Virginia and Pennsylvania. Between 2019 and 2023, 57,282 crime guns recovered in its market states, from the District of Columbia up to Massachusetts, traced back to those six.
Most of Florida's own traced guns start at home, though. ATF's 2024 trace sheet for Florida counts 36,766 firearms recovered and traced in the state that year. A source state was identified for 31,407 of them, and for 24,970 the source state was Florida. That is about four in five. The guns police recovered in Florida were, overwhelmingly, last sold by Florida licensees.
The cost is not abstract. Among state prisoners who had a gun during their offense, the prison survey found, 27 percent killed someone with it.
The strongest version of the other side
A person who means to rob or kill is not going to be stopped by a form. The prison survey backs that up as far as it goes: about one in ten of these prisoners used a retail counter, and fewer than one in a hundred used a gun show. If the people committing gun crimes already avoid the checkpoint, more rules at the checkpoint land on the hunter and the homeowner. And the data on the other side has holes. ATF says so itself, on its Florida trace sheet: "The firearms selected do not constitute a random sample and should not be considered representative of the larger universe of all firearms used by criminals." Traces capture only the guns police recover and submit. The prison survey captures only the people who got caught.
All of that is fair. Now the catch, because the show doesn't sell you a clean ending.
Avoiding the counter is not the same as having another supply. The trace record shows no second factory. It shows licensed sales, then theft, straw buyers and unlicensed resellers. The "illegal market" in these documents is the legal market a few handoffs later. Whether tightening those handoffs reduces violence is a research question, and the research is mixed. A 2009 study of 54 cities by Daniel Webster and colleagues found that strong dealer oversight and state regulation of private handgun sales "were each associated with significantly lower levels of intrastate gun trafficking," and found no effect from one-gun-a-month limits. That is an association in a cross-sectional study, not proof of cause. RAND's running review of the evidence, updated in January 2026, rates the evidence that background checks decrease firearm homicides as "moderate," and rates the evidence on mass shootings "inconclusive."
Mass shootings are where the viral claim runs backward. The National Institute of Justice reported in 2022 on a database it funded, built by Jillian Peterson and James Densley, covering 172 people who carried out public mass shootings from 1966 to 2019. Where the source of the guns could be confirmed, and in about a third of cases it could not, 77 percent had bought at least some of their guns legally. Thirteen percent made illegal purchases. That does not prove checks are useless. It says those buyers cleared the standard as written.
Why you cannot check this yourself
You can read ATF's totals. You cannot read the traces. In the 2011 law funding the Justice Department, Public Law 112-55, Congress barred spending any money, that year "and in each fiscal year thereafter," to disclose the contents of the trace database to anyone outside law enforcement, prosecutors and national security agencies. The same passage makes the data "immune from legal process" and inadmissible in civil suits, with a carve-out that lets ATF publish aggregate statistics. That carve-out permits the kind of aggregate reports cited above. A second clause forbids any rule "requiring a physical inventory" of a licensed gun business. A third requires the background check system to destroy the identifying information of an approved buyer within 24 hours.
Somebody decided the gap between the store and the crime scene should stay dark. It is in the statute, in plain words.
What you can actually do
Pull your state's trace sheet. Florida's 2024 sheet is linked above, free, and it lists the top source states and the time-to-crime spread. When a candidate tells you crime guns come from somewhere else, that page is the answer for where you live.
If you own a gun, the theft table is about you. Ninety-five percent of reported stolen guns came from private owners, not stores. Lock it up, keep it out of the truck, write down the serial number, and report a theft the day you find it, whether or not your state makes you.
And when the ninety percent claim crosses your feed, ask: ninety percent of what? Ninety percent did not buy at a retail counter. Table 6, page 8. Send them the PDF.
The so-what
The claim wants you to picture two worlds, a lawful gun market here and a criminal one there, so that any rule on the first looks like punishment for the second. The federal record describes one supply chain with a licensed business at the front, then theft, straw buyers and unlicensed resellers, and a law that keeps the details out of public view.
This show will not tell you which policy fixes that, because the research has not settled it and I will not pretend otherwise. What the documents settle is narrower, and it is enough. The ninety percent figure measures where prisoners shopped. It never measured who broke the law to get a gun, and it says nothing about where the guns began.
Come back for the rest
Every number above is walked back to its page, and the pages are listed below. Subscribe to The Long Pour, the free weekly newsletter, and the next set of receipts comes to you.
More than a quarter of the state prisoners in that survey who carried a gun killed someone with it. Nearly every crime gun ATF traces has a licensed sale on record at one end and a crime scene at the other. The least a country can do is be willing to look at the distance in between.
The Receipts
Every source below was opened and saved on September 20, 2026.
- Source and Use of Firearms Involved in Crimes: Survey of Prison Inmates, 2016 (Bureau of Justice Statistics, Alper and Glaze, NCJ 251776, January 2019; accessed September 20, 2026): Tables 5 and 6, the definitions, the survey questions and the 27 percent figure.
- Survey of Prison Inmates (SPI) (Bureau of Justice Statistics, data collection page; accessed September 20, 2026): the survey's collection years, ending in 2016.
- NFCTA Volume IV, Part III: Crime Gun Tracing Updates and New Analysis (Bureau of Alcohol, Tobacco, Firearms and Explosives, trace data 2017 to 2023; accessed September 20, 2026): licensee types, purchaser and possessor, time-to-crime, in-state share.
- NFCTA Volume IV, Part IV: Firearm Trafficking Investigations (Bureau of Alcohol, Tobacco, Firearms and Explosives, cases 2017 to 2021, traces 2019 to 2023; accessed September 20, 2026): trafficking channels, guns per case, the I-95 Pipeline.
- NFCTA Volume IV, Part II: Firearm Thefts and Losses Updates and New Analysis (Bureau of Alcohol, Tobacco, Firearms and Explosives, theft data 2019 to 2023; accessed September 20, 2026): stolen-gun totals, state rates and the theft-reporting footnote.
- Firearms Trace Data: Florida, 2024 (Bureau of Alcohol, Tobacco, Firearms and Explosives, last updated July 24, 2025; accessed September 20, 2026): Florida traces, source states and ATF's trace-data disclaimer.
- Consolidated and Further Continuing Appropriations Act, 2012 (Congress, Public Law 112-55, 125 Stat. 609 to 610 and section 511, approved November 18, 2011; accessed September 20, 2026): the trace-data, inventory and 24-hour provisions.
- Effects of State-Level Firearm Seller Accountability Policies on Firearm Trafficking (Webster, Vernick and Bulzacchelli, Journal of Urban Health, 2009, abstract; accessed September 20, 2026): what Daniel Webster and colleagues found in 54 cities.
- The Effects of Background Checks (RAND, Gun Policy in America, updated January 29, 2026; accessed September 20, 2026): RAND's evidence ratings for homicides and mass shootings.
- Public Mass Shootings: Database Amasses Details of a Half Century of U.S. Mass Shootings with Firearms (National Institute of Justice, February 3, 2022; accessed September 20, 2026): what Jillian Peterson and James Densley's database found on how mass shooters got guns.