ATF Crime Gun Trace Data: The Federal Dataset the Tiahrt Amendment Tried to Hide

Every time a law enforcement agency recovers a firearm at a crime scene and wants to know where it came from, it submits a trace request to the ATF's National Tracing Center. The NTC processes more than 500,000 of these requests every year, reconstructing the chain of commerce from factory to dealer to first retail sale to crime scene. The resulting database is the most complete record of how guns move from lawful commerce into criminal use that the federal government has ever assembled. Congress has spent more than twenty years making sure the public cannot see most of it.

This article covers how ATF crime gun tracing works, what the published aggregated data reveals, how the Tiahrt Amendment constrains public access, what the FFL directory and AFMER manufacturing data add to the picture, and how investigators and researchers use these datasets together to document the iron pipeline, straw purchasing patterns, and the ghost gun surge.

How gun tracing works: the National Tracing Center

The National Tracing Center, located in Martinsburg, West Virginia, is the sole federal repository for firearm transaction records. Its core function is tracing — reconstructing the chain of commerce for any firearm recovered at a crime scene or in a law enforcement investigation.

When a law enforcement agency recovers a firearm, it can submit a trace request through eTrace, ATF's web-based submission portal. The agent enters the firearm's make, model, caliber, and serial number. The NTC then works backward through the distribution chain: it contacts the manufacturer to identify the original distributor, contacts the distributor to identify the licensed dealer, and contacts the dealer to obtain the out-of-business records or bound book entries that identify the first retail buyer. That buyer is the “first purchaser of record” — not necessarily the person who used the gun in a crime, but the starting point for further investigation.

Dealers are required under federal law to maintain acquisition and disposition records — the bound book — for every firearm they receive and transfer. When a dealer goes out of business, those records are forwarded to the NTC, which digitizes and indexes them. The NTC holds approximately 1 billion out-of-business dealer records, a cumulative archive of decades of retail firearm transactions. Active dealers respond to trace requests directly; their records are not held centrally, which adds time to traces involving recently purchased firearms still in the dealer's active records.

The NTC can complete most traces within 24 to 48 hours when records are digitized. Handwritten or microfilm records from older transactions can take substantially longer. The trace success rate — the fraction of submitted trace requests that result in a complete chain of custody from manufacturer to first retail buyer — is approximately 70 to 75 percent. Untraceable guns include firearms manufactured before serial number requirements took effect, imported guns with insufficient documentation, and, increasingly, privately made firearms that were never serialized in the first place.

eTrace: the submission system

eTrace is the secure web portal that law enforcement agencies use to submit trace requests to the NTC. Access requires agency registration and is available to federal, state, local, tribal, and international law enforcement. As of the mid-2020s, approximately 7,000 domestic and 6,000 international agencies are registered eTrace users.

eTrace does more than accept submissions. It returns trace results electronically and allows agencies to search their own historical trace records, export data for local analysis, and query ATF's multiple-sales database — a record of transactions in which a single buyer purchased two or more handguns from the same dealer within five business days. Multiple-sale records are one of the few proactive firearms intelligence tools ATF has; they flag potential straw purchasing before a gun reaches a crime scene.

International agencies participating in eTrace — primarily in Mexico, Canada, and Central America — submit traces on firearms recovered in their jurisdictions. The resulting data has been used to document U.S. gun trafficking into Mexico; ATF analyses of Mexican crime gun traces have consistently found that 70 to 90 percent of traceable firearms recovered by Mexican law enforcement originate from U.S. dealers, primarily in border states.

The Tiahrt Amendment: why the database stays closed

The Tiahrt Amendment is an appropriations rider first attached to the Commerce, Justice, Science appropriations bill in fiscal year 2003, sponsored by then-Representative Todd Tiahrt of Kansas. It has been renewed in some form in every subsequent appropriations cycle. Its core provisions prohibit ATF from releasing individual firearm trace data to the general public and from sharing trace data with anyone outside of law enforcement for criminal justice purposes.

Before Tiahrt, cities including Chicago, New York, and Los Angeles had successfully obtained trace data through litigation and used it to identify which specific dealers supplied the highest volumes of crime guns. The cities were building civil suits against gun manufacturers and dealers based on the theory that dealers who supply disproportionate numbers of crime guns bear civil liability for the foreseeable downstream harm. Tiahrt was a direct response to those lawsuits — drafted with support from the NRA and the firearms industry — to cut off the data pipeline that made them viable.

The Amendment's practical effect: law enforcement agencies can access trace data for their own investigations. The NTC can share data with other law enforcement agencies investigating specific crimes. But ATF cannot publish dealer-level trace data, cannot share individual trace records with researchers, journalists, or the public, and cannot release data in a form that identifies which specific dealers or geographic areas supply the most crime guns. Only aggregated, state-level data is publicly available.

The Amendment also restricts how the National Instant Criminal Background Check System (NICS) data can be retained: records of approved firearms purchases must be destroyed within 24 hours, preventing ATF from building a centralized registry of who owns what firearms. Combined with the trace data restriction, this means the federal government has robust data on which guns were recovered at crime scenes but is legally prohibited from publicly connecting that data to which dealers sold them.

The published data: what ATF does release

ATF publishes four data products that, while far less granular than the underlying trace database, are analytically useful when used together.

Crime Gun Trace Reports

ATF publishes annual Crime Gun Trace Reports summarizing trace activity at the state and, in recent editions, the city level. The reports include total traces by state, the top source states for crime guns recovered in each state, time-to-crime distributions, and data on the types of firearms most frequently traced. Historical reports from 2000 through the present are available as PDFs on atf.gov; some years also include machine-readable tabular data.

The reports show, at the state level, what fraction of crime guns recovered in that state were originally sold in that state versus imported from other states. This “in-state sourcing rate” is the primary quantitative measure of how much a state's gun laws affect the supply of crime guns within its borders. States with strict purchase laws — requiring background checks on private sales, handgun permits, or waiting periods — tend to have lower in-state sourcing rates, meaning more of their crime guns come from other states. States with permissive laws tend to supply guns to stricter neighboring states.

Firearms Trace Data by State

ATF also publishes annual state-level trace data tables, available on atf.gov, showing total traces recovered in each state by year, the top ten source states for guns recovered in each state, and breakdowns by firearm type and time-to-crime bucket. These tables are the most directly machine-readable portion of the public ATF trace data and are the starting point for iron pipeline analysis.

FFL directory

The Federal Firearms Licensee directory is published monthly on atf.gov as a set of pipe-delimited text files, one per state. It contains every active licensee: name, trade name, address, city, state, ZIP code, and license type code. License type codes relevant to retail commerce are:

• 01 — Dealer in firearms other than destructive devices (standard gun store)
• 02 — Pawnbroker dealing in firearms
• 06 — Manufacturer of firearms other than destructive devices
• 07 — Manufacturer of destructive devices, ammunition for destructive devices, or armor-piercing ammunition
• 08 — Importer of firearms other than destructive devices or ammunition
• 09 — Dealer in destructive devices
• 10 — Importer of destructive devices, ammunition for destructive devices, or armor-piercing ammunition
• 11 — Manufacturer of ammunition for firearms other than destructive devices or armor-piercing ammunition

The FFL directory is the foundation for dealer density mapping. Researchers compute licensed dealers per capita by state or county to study whether dealer density correlates with gun violence rates, crime gun sourcing volumes, or background check refusal rates. The FFL directory does not contain the inspection history, compliance record, or trace volume for any dealer — all of that is blocked by Tiahrt.

AFMER: Annual Firearms Manufacturers and Export Report

The Annual Firearms Manufacturers and Export Report (AFMER) is published yearly and covers domestic firearms production by manufacturer and firearm type. The data shows total units manufactured in each category: pistols, revolvers, rifles, shotguns, machine guns, silencers/suppressors, and destructive devices. AFMER data runs from the 1980s to the present on atf.gov.

AFMER documents the dominant production trends of the last two decades. Pistol production surpassed rifle production in the mid-2000s and has remained the largest single category. The AR-platform rifle surge began after the 2004 expiration of the federal assault weapons ban: rifle production roughly tripled between 2005 and 2015, driven almost entirely by AR-style sporting rifles. Suppressor production has grown roughly tenfold since 2010 as the NFA transfer market expanded and suppressor companies proliferated; the category counted fewer than 250,000 registered suppressors in 2010 and surpassed 2.7 million by the mid-2020s.

AFMER data can be compared to trace data to identify which manufacturers appear disproportionately in crime gun recoveries relative to their production volumes — but only at the aggregate level, because the trace database that would enable manufacturer-level analysis is not publicly available. The published Crime Gun Trace Reports do include a table of the top manufacturers traced nationally, which enables a rough production-to-trace ratio. Glock and Smith & Wesson consistently lead both production and trace volumes; their elevated trace shares are largely proportional to their market dominance.

Time-to-crime: the straw purchase signal

Time-to-crime is the interval between a firearm's first retail sale and its recovery at a crime scene. It is one of the most analytically powerful variables in the ATF trace data because it distinguishes different pathways from commerce to crime.

A short time-to-crime — defined in ATF reports as under three years, and particularly flagged when under one year — is associated with straw purchasing: the practice of using a person with a clean record to buy a gun on behalf of someone who cannot pass a background check. Straw purchases move guns from retail commerce directly to prohibited persons, compressing the chain of custody. Nationally, roughly 20 to 25 percent of traced crime guns have a time-to-crime under three years; in high-trafficking corridors like the iron pipeline states, that share rises sharply.

A long time-to-crime — ten years or more — typically indicates a gun that moved from its first buyer through theft, private resale, or extended personal use before reaching the crime scene. These guns are much harder to connect to a specific diversion event. Nationally, the median time-to-crime for traced firearms has been running around seven to eight years, but the distribution is bimodal: a sharp spike at under three years (trafficking) and a long tail extending past twenty years (theft and secondary market circulation).

ATF publishes the percentage of traces with time-to-crime under three years by state, which functions as a rough state-level trafficking indicator. States with high short-time-to-crime percentages tend to be either high-trafficking source states or states recovering guns from active trafficking pipelines. The metric is imperfect — it reflects only traced guns, not all crime guns, and the traceability rate varies by state — but it is one of the few trafficking-specific signals in public federal data.

The iron pipeline

The “iron pipeline” is the colloquial term for the trafficking routes that move firearms from permissive-law states in the South and Southeast to strict-law states in the Northeast and Mid-Atlantic. The term originated in New York law enforcement but is now used by ATF and academic researchers to describe documented interstate trafficking flows.

The ATF's published state-level trace data documents the pipeline quantitatively. For New York State, the top source states for crime guns have consistently included Georgia, Florida, Virginia, South Carolina, North Carolina, and Pennsylvania. The pattern is stable across years: Georgia and South Carolina alone account for double-digit percentages of out-of-state crime guns recovered in New York. For Washington, D.C., Virginia has historically been the dominant source state. For Massachusetts, New Hampshire — which borders Massachusetts and has permissive gun laws — has supplied crime guns at rates dramatically out of proportion to its population.

The pipeline operates because gun laws are only as restrictive as the weakest link in the chain. New York requires a license to purchase a handgun, bans assault weapons, and mandates background checks on all sales. Georgia requires no background check on private sales, has no waiting period, and issues no handgun license. A trafficker who purchases firearms at a Georgia gun show or through Georgia private sales and drives them north on I-95 exploits that regulatory gap at every step. The ATF data shows the result: the I-95 corridor is the dominant geographic footprint of iron pipeline trafficking.

Research using the publicly available ATF state trace data has quantified the relationship between state gun law strength and crime gun importation rates. States with strong gun laws import a larger fraction of their crime guns from other states, while states with permissive laws export more crime guns. The relationship is not uniform — population density, geography, and border configuration all matter — but it is statistically robust across multiple studies and multiple years of ATF data.

The ghost gun problem

Privately made firearms — ATF's term for what are colloquially called ghost guns — are firearms assembled from parts kits or 3D-printed components that were never serialized and never passed through a licensed dealer. Because they have no serial number and no first retail sale, they cannot be traced through the NTC system. They are effectively invisible to the crime gun tracing infrastructure.

ATF began publishing data on privately made firearm recoveries in 2021. The numbers documented a steep acceleration: law enforcement submitted approximately 1,700 PMF trace requests in 2016, a figure that reached nearly 20,000 in 2021 and continued growing. Major urban police departments — Baltimore, Los Angeles, Philadelphia — reported that PMFs were accounting for 20 to 40 percent of gun recoveries in some enforcement contexts.

The PMF surge complicates iron pipeline analysis because imported ghost guns do not create a trail in either the AFMER production data or the dealer-level chain of custody. They enter the crime gun pool outside the regulated distribution system entirely. ATF's 2022 final rule redefining “firearm” to include weapon parts kits and requiring serialization of PMFs sold commercially was a direct regulatory response to this gap, but unserialzed guns already in circulation remain untraceable.

FFL dealer density: download and compute

Mapping FFL density is the most accessible form of firearms data analysis because the underlying data is public, updated monthly, and requires no special access. The Python code below downloads the current FFL directory for all U.S. states, filters to active retail dealers (license types 01 and 02), and computes dealers per 100,000 population using 2020 Census figures. The result reveals dramatic variation: states like Wyoming, Montana, and Alaska run 60 to 100 dealers per 100,000 residents; California, New Jersey, and Massachusetts run under 5. Dealer density correlates weakly with crime gun export rates in the ATF trace data but correlates more strongly with per-capita NICS background check volumes, suggesting that dealer access shapes purchasing behavior beyond what crime gun flows alone capture.

import requests
import pandas as pd

# ATF publishes the FFL directory monthly as pipe-delimited text files, one per state.
# The base URL pattern is: https://www.atf.gov/firearms/docs/undefined/[STATE]ffllist/download
# For a bulk pull, iterate over all two-letter state codes.

STATE_CODES = [
    'AL','AK','AZ','AR','CA','CO','CT','DE','FL','GA',
    'HI','ID','IL','IN','IA','KS','KY','LA','ME','MD',
    'MA','MI','MN','MS','MO','MT','NE','NV','NH','NJ',
    'NM','NY','NC','ND','OH','OK','OR','PA','RI','SC',
    'SD','TN','TX','UT','VT','VA','WA','WV','WI','WY',
    'DC','GU','PR','VI',
]

# FFL license type codes — dealers only (01 = dealer, 02 = pawnbroker)
DEALER_TYPES = {'01', '02'}

# 2020 Census state populations for per-capita calculation
STATE_POP = {
    'AL': 5024279, 'AK': 733391, 'AZ': 7151502, 'AR': 3011524,
    'CA': 39538223, 'CO': 5773714, 'CT': 3605944, 'DE': 989948,
    'FL': 21538187, 'GA': 10711908, 'HI': 1455271, 'ID': 1839106,
    'IL': 12812508, 'IN': 6785528, 'IA': 3190369, 'KS': 2937880,
    'KY': 4505836, 'LA': 4657757, 'ME': 1362359, 'MD': 6177224,
    'MA': 7029917, 'MI': 10077331, 'MN': 5706494, 'MS': 2961279,
    'MO': 6154913, 'MT': 1084225, 'NE': 1961504, 'NV': 3104614,
    'NH': 1377529, 'NJ': 9288994, 'NM': 2117522, 'NY': 20201249,
    'NC': 10439388, 'ND': 779094, 'OH': 11799448, 'OK': 3959353,
    'OR': 4237256, 'PA': 13002700, 'RI': 1097379, 'SC': 5118425,
    'SD': 886667, 'TN': 6910840, 'TX': 29145505, 'UT': 3271616,
    'VT': 643077, 'VA': 8631393, 'WA': 7705281, 'WV': 1793716,
    'WI': 5893718, 'WY': 576851, 'DC': 689545,
}

def fetch_ffl_state(state: str) -> pd.DataFrame:
    url = (
        'https://www.atf.gov/firearms/docs/undefined/'
        + state.lower() + 'ffllist/download'
    )
    resp = requests.get(url, timeout=30)
    resp.raise_for_status()
    from io import StringIO
    df = pd.read_csv(
        StringIO(resp.text), sep='|', dtype=str,
        names=[
            'region', 'dist_office', 'county', 'license_rgn_cd', 'license_dist_cd',
            'license_cnty_cd', 'license_type', 'expire_dt', 'license_name',
            'business_name', 'prem_street', 'prem_city', 'prem_state',
            'prem_zip', 'mail_street', 'mail_city', 'mail_state', 'mail_zip',
            'phone',
        ],
        on_bad_lines='skip',
    )
    df['prem_state'] = state
    return df

frames = []
for st in STATE_CODES:
    try:
        frames.append(fetch_ffl_state(st))
    except Exception as e:
        print(f'  {st}: {e}')

all_ffls = pd.concat(frames, ignore_index=True)

# Filter to active retail dealers only (type 01 and 02)
dealers = all_ffls[all_ffls['license_type'].isin(DEALER_TYPES)].copy()

# Dealers per 100k population by state
state_counts = dealers.groupby('prem_state').size().reset_index(name='dealer_count')
state_counts['population'] = state_counts['prem_state'].map(STATE_POP)
state_counts = state_counts.dropna(subset=['population'])
state_counts['dealers_per_100k'] = (
    state_counts['dealer_count'] / state_counts['population'] * 100_000
).round(1)

print(
    state_counts.sort_values('dealers_per_100k', ascending=False)
    .to_string(index=False)
)

Journalist use cases

Investigative journalists have developed a set of reproducible methodologies for the publicly available ATF data despite the Tiahrt constraints.

Iron pipeline mapping. Using the ATF state trace data tables, journalists map the dominant source states for crime guns in specific cities or states, compute in-state versus out-of-state sourcing rates, and track how those ratios change year over year. The technique has been used by The Trace, The New York Times, and the Washington Post to document how changes in state gun laws affect crime gun importation rates.

FFL density and gun violence correlation.By joining the FFL directory to county-level CDC WISQARS gun mortality data, researchers and journalists have tested whether higher dealer density in a county correlates with higher firearm mortality rates, controlling for urbanicity, poverty, and other covariates. The relationship is complicated by the distinction between dealer density in the county where violence occurs versus dealer density in the source counties, but the FFL directory makes the analysis tractable.

Manufacturer production trends. AFMER data, plotted over two decades, documents the AR-platform surge, the pistol dominance that followed it, and the suppressor market expansion. Journalists have used AFMER data to contextualize industry responses to proposed regulations and to document how production volumes changed after high-profile mass shootings.

Straw purchase hot-spots. The ATF Crime Gun Trace Reports' short-time-to-crime tables, published by state and for some cities, allow journalists to identify jurisdictions with unusually high concentrations of recently purchased crime guns — a signal for active trafficking networks. Pairing that with the FFL directory, journalists can identify which dealers are geographically proximate to identified trafficking corridors and submit FOIA requests for inspection records from those dealers.

Cross-reference datasets

ATF crime gun data produces its strongest analytical results when combined with other federal datasets that measure adjacent phenomena.

The FBI's NIBRS crime database documents violent crimes involving firearms at the incident level. NIBRS weapon type codes distinguish handguns, rifles, shotguns, and other firearms; pairing NIBRS firearm-involved offense rates by state with ATF trace rates by state provides a fuller picture of how gun availability connects to gun violence across jurisdictions. NIBRS is the only federal dataset that produces jurisdiction-level firearm offense counts at a granularity comparable to the ATF state trace data.

The CDC's WISQARS (Web-based Injury Statistics Query and Reporting System) publishes firearm mortality and morbidity data by state, mechanism (homicide, suicide, unintentional, undetermined), and year. WISQARS mortality data reflects actual gun deaths, not crime gun recoveries; many suicides and accidents involve legally purchased firearms that would never generate a trace request. Comparing ATF trace volumes to WISQARS mortality rates by state highlights the difference between enforcement-visible gun violence (the trace universe) and population-level gun harm (WISQARS).

The DOJ's NICS background check data — published monthly by the FBI as the NICS Firearm Checks report, broken down by state and by sale versus permit type — documents the volume of firearm purchases flowing through licensed dealers. NICS check volumes, combined with AFMER production data and ATF trace counts, give a rough approximation of the denominator for crime gun rates: how many guns entered commerce relative to how many were later recovered at crime scenes. The ratio has historically been very small — roughly 1 to 2 percent of manufactured handguns are ever traced to a crime scene — but the distribution across dealers is highly skewed: a small fraction of dealers account for a disproportionate share of crime gun traces, a pattern documented in pre-Tiahrt data that the Amendment has since made impossible to verify at the dealer level.

NHTSA's FARS traffic fatality database and ATF trace data share a methodological structure: both are administrative records of a specific type of harmful event — fatal crashes and crime gun recoveries — that capture only the events that enter the formal reporting system. Both are subject to reporting rate variation and both require cross-referencing with population denominators to produce meaningful rates. Researchers building state-level public safety risk indices often incorporate both.

Accessing the data

The FFL directory and AFMER are available directly on atf.gov with no registration required. The FFL directory is updated monthly; AFMER is published annually, typically with a one-to-two year lag. The annual Crime Gun Trace Reports and state firearms trace data tables are on the ATF trace data page at atf.gov/resource-center/firearms-trace-data. All of these are public domain federal data products.

Individual trace records, dealer-level trace data, and any data that would identify which specific dealers supplied which crime guns remain restricted under the Tiahrt Amendment. FOIA requests for this data are denied. The only pathways to dealer-level data are through formal law enforcement channels (available only to law enforcement agencies) or through litigation that survives a Tiahrt-based government privilege assertion — a high bar that few cases have cleared.

The practical effect is that the federal government has built, and is actively operating, a sophisticated surveillance system that maps the flow of firearms from commerce to crime at a level of detail that has no parallel in federal law enforcement data. It runs on public funding. And most of what it produces is unavailable to the public, the press, and the research community by explicit congressional mandate. The aggregated state-level data, the FFL directory, and AFMER are what remains after Tiahrt — the residue of a dataset that, if public, would be one of the most consequential law enforcement databases in federal transparency history.