Is there a national US parcel database?

No. There is no single database that holds every US parcel. Our own catalogue is the proof: of 272 connected sources on 2026-09-11, not one is a national US parcel source, nine states have none at all, and the 158 sources carrying no state label are almost all foreign registers. What "national" would have to mean A national US parcel database would have to return a parcel record for any of the 3,143 counties in the country, from one place, with the same fields each time. Nothing we have connected does that, and we have looked at 306 sources to find out. The honest answer to the question is no, and the cheapest way to show it is to publish our own inventory rather than describe it. Every figure below comes from one public query against our data catalogue, pulled on 2026-09-11 (the response stamped itself generatedAt 2026-09-11T12:21:17Z). The coverage grades in the section on uneven states come from our internal geography record, measured 2026-09-14. Those are two different measurements on two different bases; they are reported separately here and never added together. 158 sources with no state is not a national layer As of 2026-09-11, our catalogue listed 306 sources, of which 272 were connected — the pipeline can query them today. The remaining 34 are listed too: 31 assessed and rejected, each carrying the written reason we turned it down, one in progress, one blocked by its publisher, one planned. Of those 272 connected sources, 158 carry no state value at all. That looks like the shape of a national layer, and it is not one. A missing state field means the source is not organised by US state — usually because it is not American. Twenty of the 158 are parcel or transaction sources, the only kind that could answer a parcel question. Nineteen of those twenty are not American. The catalogue names them as HM Land Registry, the French DVF and cadastre, five Spanish cadastres, and the Dutch, German, Swiss, Czech, Irish, Norwegian, Danish and Finnish registers. We do not present those countries as covered: our geography record, measured 2026-09-14, grades the United Kingdom, the Netherlands, Switzerland, Czechia, Norway, Denmark, Ireland and Finland as code only — a connector exists, production has never once seen it answer. The twentieth is American, and it is an index of three county recorders. Three counties, out of 3,143. The catalogue states the conclusion in its own words: there is no national US parcel source in this catalogue, and nothing in that bucket reaches a county in one of the nine states with no coverage. Where the map ends Nine states had no connected source of any kind on 2026-09-11: Alabama, Delaware, Hawaii, Iowa, Kansas, Kentucky, Mississippi, South Carolina, South Dakota. That is 593 of the country's 3,143 counties. Seven of the nine hold a rejected source with a written reason. Kentucky and South Carolina have no catalogue row at all, because nobody here has assessed them yet. Connected coverage stands at 41 states plus the District of Columbia — 42 jurisdiction codes in the catalogue, which is the same fact stated two ways, not a contradiction. Inside the covered states, coverage is not uniform either "Connected" is not "complete," and this is where a national claim usually breaks quietly. From our geography record as of 2026-09-14: a full cadastral sweep exists in Florida and Texas only. A statewide parcel roll exists in 19 states. The remaining 23 registered states are listed counties only — Georgia is four counties, Arizona is Maricopa and Pima, California is 11 counties of 58. As of 2026-09-11, fifteen states rest on exactly one connected source; the deepest states carry eight. A count of sources is not a measure of depth, and neither is a count of states. The fields are not uniform where the data does exist Of the 272 connected sources, 66 carry a field-level disclosure matrix — whether that source returns a parcel id, an address, a land use, an assessed value, a building area, a sale date, a sale price. The other 206 are lookups that answer about one property at a time and publish no such matrix. By kind, the 272 split into 201 lookups, 66 enumerations and 5 signals; the 83,088,677 records in the catalogue are the count carried by those 66 enumerations, and they are parcels — not transactions, and not distinct buildings. The most important number in the disclosure matrix is the smallest. 24 connected sources disclose a recorded sale price, across 19 states, all as of 2026-09-11. Recorded price is the thinnest layer in American property data. Texas is the clearest case, and the catalogue says it in the source's own words: "Texas is a non-disclosure state: the appraisal roll carries no sale price and no sale date." What we do not publish about our own catalogue We do not publish a refresh cadence, because we do not hold one. The only per-source date in the catalogue, measuredAt, exists on 97 of the 306 rows and 66 of the 272 connected ones, and every stamp reads 2026-08-13, -15 or -22. It records when we last measured that source, not when the publisher last refreshed it. Licence and redistribution rights are recorded for exactly 1 of the 272 connected sources, so we do not make a blanket statement about what each licence permits. What we will send you instead We will send the full source list, including the nine states where we hold nothing, before you sign anything.