Data sources and limitations
A search tool on a legal-adjacent topic owes you a plain statement of what is in the index, what is not, and how old it is.
Primary source
USPTO Trademark Annual Register (TRTYRAP), 1884-01-01 to 2025-12-31, 91 annual XML volumes, 11.50 GiB, obtained through a public-domain mirror of the USPTO bulk-data product; plus the daily incremental product (TRTDXFAP), 307 files covering 2026-01-01 to 2026-11-03. The authoritative source for any individual record is the USPTO itself, not us.
Daily files are fetched and merged into the index, but the publish step is run by hand rather than on a schedule, so the index can lag the register by a few days. The build date on this page is the honest answer to "how current is this?".
What is in the index
13,993,768 case-file entries were parsed from the 91 annual volumes, then 307 daily files were merged on top of that base. 4,502,862 live and unknown-status marks with searchable text are published across 47,859 record shards and 1,296 n-gram shards (49,158 compressed objects, 244.7 MiB). All 144 observed status codes were audited and mapped.
- Coverage: 1884-01-01 to 2026-11-03 (annual register plus daily incremental files).
- Index built: 2026-11-04.
- Fields per record: serial, mark, registration number, class, filing date, registration date, owner, goods/services, status code, status date.
- Status classification basis: USPTO cons181 status definitions, the USPTO status-code reference, and TM5.
What is not in the index yet
We would rather state the gaps than let you assume they are covered.
- Dead and cancelled historical marks are parsed but not yet searchable.
- Daily files are merged by an off-line job that is started by hand; filings from the last few days may be missing.
- Goods/services class overlap and owner-based weighting are not in the risk label yet.
- Design marks / figurative elements, images and logo similarity are not evaluated; this is a text-mark index.
- Non-US registers (WIPO, EUIPO, state registers, common-law use) are not covered.
Accuracy and corrections
Bulk-data transcription and status classification can be wrong. If you find a record that disagrees with TSDR, TSDR wins. Please report it through the corrections channel so we can fix the mapping or re-ingest the source.
Corrections go to the corrections address on the contact page. Tell us the serial number, the page you were on and the official TSDR record. Confirmed mapping errors are fixed in the next index build; the build date is shown on this page, so you can see whether your report has landed.