Methodology
Where these figures come from, how they are built, and what they cannot tell you.
The source
Everything here derives from the FCC's Universal Licensing System bulk data
downloads — the same public archives the Commission publishes at
data.fcc.gov/download/pub/uls. No figure on this site is
hand-entered, estimated, or sourced from a third party. Nothing is scraped
from a web page.
Two feeds are used. A weekly full archive carries every Amateur Radio
licence and application record the FCC holds (services HA and HV). A daily delta archive
carries just the records that changed the previous day. The current set was
built from licence data published Sep 13, 2026 and
application data published Sep 19, 2026.
Two pipelines, and why it matters which one you are reading
The weekly archive rebuilds the licence tables from scratch on every load. Counts, state breakdowns, growth history, call sign series, licensing eras and the expiry horizon are all derived from that snapshot, so they are complete regardless of what happened on any individual day.
Daily Transaction Flow is the exception. It is the only page built from the daily deltas, accumulated into a change log one day at a time. That makes it the only page that can have a hole in it — and holes there are permanent, because the FCC overwrites each weekday file the following week and retains no history. A licence missed by the daily feed still appears in the next weekly snapshot and in every total on this site; what is lost is only the ability to say which day it arrived.
Dates are offset by one day
A daily archive is compiled overnight and published the following morning, so the file published on a Tuesday describes Monday's activity. Dates shown on this site are transaction dates — the day the activity happened — not the day the FCC published the file describing it.
How to read a blank
A value of 0 and a blank — mean different things,
and the distinction is enforced rather than left to chance.
- 0 — the day was measured and the count was zero.
- — — the day was not measured, so no honest number exists.
Of the 9 days currently in the daily record, 0 are blank. Every blank carries a machine-readable reason, and the reason is published alongside the data so it can be checked rather than taken on trust. Coercing those to zero would make an unmeasured day indistinguishable from a genuinely quiet one, and would quietly corrupt any average drawn from the series.
Every figure declares what kind of claim it is
- observed — counted directly from FCC records.
- projected — derived from observed data by a stated rule, such as when currently active licences will reach the end of their ten-year term.
- unavailable — the FCC does not publish what would be needed to answer.
Where a period is affected by something outside the data — an upstream outage, a holiday, a day the FCC published nothing — an annotation records it once, and every page that shows that period displays the same explanation. There is 1 such annotation in the current set.
What is deliberately not published
ULS records carry licensee names, street addresses, FRNs and other personal fields. None of them reach this site. The published files are aggregates only, and the publishing step is blocked automatically if a prohibited field is detected in the output.
Geographic figures are drawn from the licensee's mailing address, which is not necessarily where a station operates — Amateur Radio licenses authorize portable, mobile, and remote operations nationwide. Any group smaller than 25 licences is withheld rather than published, to avoid identifying individuals in sparsely licensed areas. That is why the state table sums to 823,359 rather than the official 823,427 — a difference of 68.
What is counted
A figure described as a license count is a count of active license authorizations (call signs), not of individual operators. While individuals may hold only one primary operator license, clubs, military recreation stations, and trustees may hold additional station licenses.
Verification before publication
Each build is checked against a published file contract before it is released: every file is validated against a schema, series are checked for gaps, counts are reconciled against the row totals the FCC declares in its own archives, and the privacy rules above are re-tested. If any check fails, publication stops and the previous set stays live. The current build parsed 1,695,551 of 1,695,551 declared licence rows and 3,415,177 of 3,415,177 declared application rows.
Corrections
If a figure here looks wrong, it may well be. The underlying archives are public and the derivation is documented, so any number on this site can be reproduced from source. Corrections are welcome via the contact form.