Operational Intelligence · 2026 Report
What this is
In June this project published its first long report, The Two-Queue Regime, on 2,462 approval records. Since then eight shorter editions have followed the queue a few weeks at a time: the Form 1 freeze, the split, the break, the drain, the closing of the form gap and the two-business-day queue. Each was right about its own few weeks. None of them could show the whole year on one scale, because the year was not over.
This report does. It covers 4,150 e-filed approvals reported by NFA filers, 3,936 of them approved in 2026, snapshotted September 25, 2026. Community data, not official ATF records, except where a figure is labelled as ATF's own. Every median carries a 95% confidence interval in square brackets, and every lane is reported separately: Form 1 and Form 4, individual and trust, never blended.
For the wait right now, see the live wait times or estimate your own filing. This page is the record of how the queue got there.
The year in one chart
Nine Months of NFA Wait Times, Lane by Lane
Median wait in days from filing to approval, by the month the approval landed, for each of the four e-file lanes. Form 4 · Individual stays in single digits or low teens all year. The other three lanes climb from January, peak between June and July, and fall back below a week by September.
Months with fewer than 15 approvals in a lane are left blank.
January was the last ordinary month. All four lanes ran a median of 10 to 13 days: Form 4 Individual 11 [10–12], Form 4 Trust 10 [9–13], Form 1 Individual 13 [12–14] and Form 1 Trust 12 [9–14]. On January 1 the National Firearms Act tax on Form 1 and Form 4 had fallen to $0, and the filings that change released were already in the queue.
What followed happened in three lanes out of four.
- –Form 1 climbed for five months. Form 1 Individual went 35 days in February, 52 in March, 61 in April, 64 in May and 69 in June. Form 1 Trust ran within a few days of it until June and then kept going, to a peak of 83 [72–84] in July.
- –Form 4 Trust stepped up once and stayed there. It jumped to 27 days in February and sat between 26 and 31 for six months.
- –Form 4 Individual barely noticed. It never rose above its January level, and from March through July it ran at 5 to 8 days.
Then the queue broke. Form 1 Individual fell from 48 days in July to 5 [4–6] in August. Form 1 Trust and Form 4 Trust followed a few weeks behind. Over the 30 days ending September 25 the four lanes ran 2 [2–3] for Form 4 Individual, 4 [3–5] for Form 4 Trust, 3 [2–4] for Form 1 Individual and 3 [2–4] for Form 1 Trust. All four are faster than January on intervals that do not overlap. Counted in business days, individual filings placed between August 11 and September 11 cleared in a median of 2, and 98% within 5.
⚠️ The September figures rest on 27 to 54 approvals per lane. They are the thinnest numbers in this report, for a reason set out in the last section of this report.
When you filed mattered more than anything you controlled
A median by approval month answers ATF's question: how old was the work it cleared this month. A filer asks a different one: I filed in March, so what happened to me? That needs the same data grouped by the month the form was filed.
What Your Filing Month Cost You
Median wait in days for filings placed in each month of 2026, by lane. This groups people by when they filed, not when they heard back, so it answers the question a filer actually asks. April was the worst month to file a Form 1 on either owner type; August was the best.
September filings are excluded: some are still waiting, so their median would be a floor, not a result.
| Filed in 2026 | Form 1 · Individual | Form 1 · Trust | Form 4 · Individual | Form 4 · Trust |
|---|---|---|---|---|
| January | 28 days · 52% | 39 days · 43% | 12 days · 98% | 23 days · 82% |
| February | 62 days · 1% | 65 days · 8% | 6 days · 97% | 26 days · 77% |
| March | 64 days · 4% | 70 days · 4% | 4 days · 96% | 27 days · 65% |
| April | 77 days · 7% | 83 days · 6% | 6 days · 97% | 27 days · 65% |
| May | 57 days · 13% | 75 days · 2% | 6 days · 98% | 29 days · 53% |
| June | 36 days · 30% | 52 days · 14% | 5 days · 99% | 31 days · 43% |
| July | 14 days · 100% | 25 days · 72% | 7 days · 100% | 24 days · 77% |
| August | 3 days · 100% | 6 days · 100% | 4 days · 100% | 8 days · 100% |
April was the worst month of the year to file a Form 1. Individual filings placed in April waited a median of 77 days [73–79] on 98 approvals, and only 7% were approved within a month. Trust filings placed in April waited 83 [82–84] on 84. February and March were barely better. A Form 1 filed in August, four months later, cleared in 3 days [3–4] for an individual and 6 [5–8] for a trust.
That is a difference of more than two months between two identical forms, decided by nothing but the date on them. Nothing a filer could choose in 2026, not the form, not individual or trust, not the day of the week, moved the wait as far as the month did.
The Chance of Hearing Back Within a Month
Share of each filing month's approvals that arrived within 30 days, by lane. A Form 1 filed in February, March or April had well under a one-in-ten chance. One filed in August had a certainty.
September filings excluded, as above.
Form 4 Individual never had a bad month. Every filing month of 2026 cleared at a median of 4 to 12 days, with at least 95% approved inside 30. Its worst month was January, the first rush of $0 filings, at 12 days [12–12] on 306 approvals. The surge happened entirely in the other three lanes.
Form 4 Trust was slow and steady rather than stuck. Its filing months ran 23 to 31 days from January through July, and the share approved within a month slid from 82% to 43% before recovering. It never froze the way Form 1 did, and it never got fast either, until August.
The queue spent four months clearing old work
How Much of Each Month Was Old Work
Share of each month's approvals that had been waiting more than 60 days, all lanes together. From April through July, between a quarter and two fifths of everything approved was two-month-old work. By September there was none.
In January and February no approval in this dataset had waited more than 60 days. From April through July, between a quarter and two fifths of everything approved had: 27% in April, 42% in May, 38% in June and 28% in July. Almost all of it was Form 1. In August that share fell to 7%, and in September it is zero.
That is what a backlog looks like from the approval side. ATF was not slow on every form for those four months. It was working through a stack of Form 1 filings placed during the winter surge, in order, while Form 4 Individual kept moving at its usual pace beside it.
The trust penalty was two different things
Every edition this year has reported a trust penalty. Read across nine months, it turns out not to be one thing.
The Trust Penalty Was Two Different Things
Extra days a trust filing waited over an individual one, median against median, by approval month. On Form 4 the gap opened in February and held near three weeks for six months. On Form 1 there was no gap while the queue was frozen; it opened only as the queue began to move, and trusts were released last.
A gap counts as real only where the two 95% intervals do not overlap; that is true for Form 4 February to August and Form 1 June to August.
On Form 4 it was a steady tax. In every month from February through August, Form 4 trust filings waited 17 to 23 days longer than individual ones, median against median, on intervals that never came close to touching. At its widest, a trust waited more than five times as long: 27 days [26–29] against 5 [4–6] in March. That gap did not grow with the Form 1 backlog and did not shrink when the Form 1 backlog drained. It behaved like a separate, slower process that happened to sit beside a fast one.
On Form 1 it did not exist while the queue was backed up. From February through May, Form 1 Trust and Form 1 Individual ran within four days of each other every month, on overlapping intervals: in April, 61 [59–65] against 61 [60–62]. When nothing was moving, trust and individual Form 1s were stuck together.
It appeared only on the way out. As the Form 1 queue began to clear, individual filings came out first. In July Form 1 Individual fell to 48 days while Form 1 Trust rose to its peak of 83, a gap of 35. In August the gap was 25: 5 days [4–6] against 30 [19–44]. Grouped by filing month, trusts placed in May waited 75 days against 57 for individuals. The trust penalty on Form 1 was a queue-release effect, not a standing rule.
In September both gaps are inside the margin of error: Form 4 Trust at 4 [2–5] against 2 [2–3], and Form 1 Trust at 3 [2–3] against 2 [2–3]. That is a real change from every earlier month, and it rests on 28 and 21 trust approvals. It is the finding most in need of another month of data.
The slowest months were not the least certain
A long wait is hard. An unpredictable one is harder, because it cannot be planned around. The two did not arrive together in 2026.
The Slowest Months Were Not the Least Certain
The range covering the middle half of Form 1 waits (the 25th to 75th percentile), in days, by approval month. While the queue was frozen almost everyone waited about the same two months. The widest spread came in July, at the turn, when one filer might hear back in three weeks and the next wait nearly three months.
While the Form 1 queue was backed up, its waits were long and tightly bunched. From March through May, the middle half of Form 1 Individual approvals fell within a 6-day range each month: in May, between 62 and 68 days. Almost everyone waited about two months, and knew roughly how long it would be.
The widest spread came at the turn. In July, the middle half of Form 1 Individual waits spanned 42 days, and the range from the 10th to the 90th percentile spanned 62: one filer heard back in three weeks and another waited nearly three months, in the same month. Form 1 Trust hit its own widest point a month later, with the middle half spanning 55 days in August. The most uncertain time to be waiting on a Form 1 was not the slowest time. It was the handful of weeks when the queue was changing speed.
The queue kept its order throughout
A filing-date estimate is only worth having if the queue runs roughly first-in, first-out. If ATF picked filings out of order, the date on a form would say little about when it will clear.
It did not. The FIFO audit in the Lab compares every pair of filings in a lane placed at least two weeks apart, and asks how often the later one was approved first. Read at five points across the year, over the 180 days before each, that rate ran between 0.5% and 1.8% in March and between 1.3% and 3.6% in September, in every lane. The small rise late in the year is the queue speeding up: with a two-day wait, any filing still pending from earlier weeks is overtaken by everything behind it. Through the backlog and its drain, the order held.
Weekend approvals stopped before the drain, not during it
Weekend Approvals Stopped Before the Drain
Share of each month's approvals dated on a Saturday or Sunday. Form 1 ran a fifth to a third of its approvals on weekends from January through May, then dropped to under a tenth in June. The backlog drained in July and August without them.
From study 05 in the Lab. Months with fewer than 30 approvals on a form are left out.
From January through May, between 21% and 34% of Form 1 approvals in this dataset carried a Saturday or Sunday date, against 12% to 18% for Form 4. An operation spread evenly across seven days would put 29% of its work on weekends. In June the Form 1 share fell to 8%, and it has stayed between 8% and 14% since, in line with Form 4.
The timing is the finding. If the Form 1 backlog had been cleared by weekend overtime, weekend approvals should have peaked in July and August, when it drained. Instead they stopped a month before. Whatever cleared the backlog, and ATF's own figures below say it was both more capacity and fewer new filings, it ran on weekdays.
An outside check, and a test we are setting ourselves
ATF publishes a small set of NFA processing figures and overwrites them every month. This project has recorded each reading since June, because the page keeps no archive.
Our Sample Against ATF's Own July Averages
ATF's published average processing time for eForms finalized in July 2026, beside the mean wait in this dataset for approvals in the same month. Means against averages, lane for lane; ATF does not split Form 1 by owner type, so Form 1 is pooled on both sides.
Our August Form 1 mean is 18.7 days (n=295). ATF's August table, when posted, is the test.
The sample agrees with ATF where the two can be compared. For July 2026, ATF published average processing times of 57 days for eForm 1, 33 for Form 4 Trust and 9 for Form 4 Individual. The mean wait in this dataset for approvals in July was 54.0, 32.1 and 9.4 days, on 318, 89 and 143 approvals. That is within three days on every lane ATF publishes. ATF's median for individual eForm 4 in July was 8 days; ours was 8 [7–9].
⛔ These are means against averages, on purpose. Every other figure in this report is a median, and setting a median against ATF's average would manufacture a disagreement out of two definitions.
ATF's own ledger shows how the Form 1 backlog cleared. Between its June 1 and August 5 readings, ATF's count of outstanding Form 1 applications fell from 52,870 to 18,651, a drop of 65%. Form 4's fell 14% over the same period. Two things moved at once: Form 1 intake fell by roughly 41% against its first-five-month pace, and Form 1 throughput rose by roughly 60%. Neither alone accounts for the drop. Even after it, Form 1 carried about twice Form 4's proportional backlog, 9.2% of the year's submissions still outstanding against 4.3%, although Form 4 runs more than five times the volume.
The test. ATF had not published its August figures when this report went out. So here are ours first. The mean wait in this dataset for approvals in August 2026 was:
- –Form 1: 18.7 days, on 295 approvals (54.0 in July)
- –Form 4 Trust: 20.7 days, on 109 approvals (32.1 in July)
- –Form 4 Individual: 5.4 days, on 96 approvals (9.4 in July)
If ATF's August averages land within about five days of those, the sample has tracked ATF through the fastest month-to-month change of the year. If they do not, that is worth knowing too, and the next edition will say so either way. One caution applies to both sides: ATF averages every application it finalized, including any disapproved or withdrawn, while this dataset holds approvals only.
Who was watching
The owner of this site has asked that its traffic be described in shares and patterns, not totals. Within that, the reading audience says something about the queue too.
Where Readers Spend Their Visits
Share of all human page views going to the live wait-time page and the estimator, by week. Together they went from a rounding error in mid-July to more than a third of everything read on the site.
The week of Sep 21 is partial.
People kept searching after the wait collapsed. Every one of the fastest-growing search phrases that brought readers here over the last two months is a wait-time question: "atf form 1 wait times", "atf form 4 wait times", "current form 4 wait times", "atf eform 4 wait times 2026", and a new cluster around SBRs, "form 1 sbr wait time 2026", which barely existed in July. The live wait-time page and the estimator went from a rounding error in mid-July to more than a third of everything read on the site. ⚠️ The site's own position in search results improved over the same weeks, so this cannot say whether more people were asking or more of them were finding the answer here. Both are probably true.
How Readers Arrive
Where each month's landing visits came from, as a share. Google's share roughly tripled between July and September. AI assistants appear nowhere, because they pass no referrer: a visit from a chatbot answer lands in “direct”.
Landing visits only. September is partial. “Internal and other” includes Facebook, Instagram and X.
Search engines took over as the way in. Google's share of landing visits roughly tripled between July and September, from 12% to 36%. Reddit, where much of this data originates, held steady at around 5%.
AI answers cite the data but do not send readers. Bing's AI answers cited this site in between 13% and 41% of responses to wait-time questions in early September, and in 58% of responses to "nfa approval". Yet visits carrying an AI assistant as their referrer run at zero: chat answers pass no referrer, so anyone who clicks through from one arrives counted as "direct".
When People Check
Share of human visits by day of the week since July 1, Eastern time. The working week carries the load, peaking midweek; Sunday draws little more than half of Wednesday.
People check on weekdays, after work. Visits peak midweek, with Wednesday the busiest day, and Sunday draws little more than half of it. The single busiest hour is 7pm Eastern, and afternoons and evenings carry two thirds of all visits. That is the rhythm of someone checking on a pending form, not of someone researching one.
What this data cannot see
Approvals Recorded per Week
Approvals in this dataset by the week they landed, all lanes. The sample thinned as the queue sped up: a fast queue leaves people less reason to post about it. Fewer records widen the uncertainty around every recent figure, which is why each one here carries its interval.
Complete weeks only.
The sample is thinning. Approvals recorded here ran mostly between 90 and 190 a week from January through May, and between 27 and 36 a week in September. The likeliest reason is the simplest: a two-day wait gives people little reason to post about it. Nothing corrects for this. The confidence intervals simply widen, which is why the September figures in this report are the least certain in it.
The year was recorded in two eras. Approvals from January through April reached this dataset weeks or months after they happened, through a retrospective import of approvals filers had already posted publicly, with a median delay of 38 to 136 days. Since June the median approval has been recorded the next day. The two eras were compared directly: in 13 of 14 lane-months where both channels had enough approvals, imported and directly reported approvals gave medians on overlapping intervals. The exception was Form 1 Individual in July, the month of the turn, where direct reports ran faster, 36 days [36–44] against 52 [46–57].
Some questions do not have enough data behind them. Item type (suppressor, SBR) has only been recorded since mid-July, which makes it a record of the fast queue only and says nothing about the year. Purchase channel and whether a Form 1 service was used have fewer than 120 answers each. None of these appear as findings here.
The sample is not ATF's queue. This dataset holds 3,936 approvals from 2026. ATF received more than 200,000 Form 1 applications and more than a million Form 4s in the first seven months of the year alone. What the sample offers is not coverage but timing: exact filing and approval dates, lane by lane, which ATF does not publish at all.
Method
Data. 4,150 e-filed Form 1 and Form 4 approvals with approval dates between January 5, 2025 and September 25, 2026, reported by NFA filers directly or collected from approvals they had posted publicly. Community data, not official ATF records.
Exclusions. The same rules every edition since August has used, applied before any figure was computed: waits outside 0 to 365 days dropped; duplicates dropped only where the form, owner type, state, filing date, approval date and masked control number all match; paper filings excluded, because the sample holds too few to describe.
Statistics. Medians throughout, by lane, with exact 95% confidence intervals from the binomial method the site uses everywhere. A lane-month with fewer than 15 approvals is not reported. Means appear only in the ATF comparison. Differences between lanes are called real only where the two intervals do not overlap, which is stricter than a direct test and chosen on purpose.
Grouping. "By approval month" groups approvals by when ATF cleared them. "By filing month" groups them by when they were filed, and is reported only for months whose filings have all cleared; September 2026 filings are excluded for that reason.
Reproducibility. Every figure in this report comes from scripts/report-analysis.ts in the NFA Watch repository, run against the extract defined in scripts/sql/report-extract.sql. The charts are locked to the September 25, 2026 snapshot and will not change. Traffic figures are shares computed from the site's own visit log, with automated traffic excluded.
For how the live figures are calculated, see the methodology. For the current wait, see the live wait times.