US ZIP Code Statistics: 40,977 Codes Measured

Most ZIP code trivia online is copied from other trivia. This page is different in one specific way: every number below is computed directly from the GeoNames US postal dataset ^[geonames], the same CC BY 4.0 file that drives the state pages and the random address generator on this site. That means each figure has a denominator you can check and a computation you can rerun — and when two figures on this site use different denominators, this page says so.

The headline numbers

One reconciliation up front, because two true numbers look like a contradiction. This page counts 911 prefixes across the 50 states and DC; other pages on this site cite 917 state-prefix pairs (a prefix serving three states counts three times) and 926 prefixes across the full file including territories and military codes. Same dataset, three denominators — the kind of footnote a statistic needs if anyone is going to cite it.

State extremes: a 29-to-1 spread

Texas edges out California by just seven codes, and Rhode Island holds the floor at 90 — a 29-to-1 spread between the extremes. Code count follows sorting territory rather than population, which is why the ordering only loosely tracks the census: Pennsylvania outranks Florida and New York despite having fewer people than either.

Density is a different league table. Massachusetts packs 48.9 codes into each of its 14 counties — the highest ratio among states — while South Dakota (5.8) and Georgia (6.1) sit at the bottom because they carry enormous county counts over modest code totals. The one true outlier is the District of Columbia: one county-equivalent, 277 codes. At the single-county level, Los Angeles County alone holds 508 codes — more than eleven entire states.

Five prefixes cross state lines

The first three digits of a ZIP code identify a sorting facility, and a sorting facility serves geography, not politics. In the 50 states and DC, exactly five prefixes serve more than one state:

These five rows are the entire exception list to "the prefix implies the state" — which is precisely what makes them valuable as test fixtures. A prefix-to-state lookup table passes every test you write against the other 906 prefixes and then rejects a real address in the capital region, where 205 covers Washington, DC plus exactly one code each in Maryland and Virginia. Add the Massachusetts 055 anomaly — two IRS codes in Andover sitting inside Vermont's range — and you have the complete set of places where prefix arithmetic betrays you.

The boundary values: 00501 and 99950

The lowest assigned ZIP code is 00501 — not a town but an IRS processing facility in Holtsville, New York, a single-organization code with its own five digits. The highest is 99950 in Ketchikan, Alaska. Both boundaries matter to anyone generating test ZIP codes: 00000 and 99999 are unassigned and always will be, which makes them the canonical invalid values, while 00501 is the awkward real code that a range check starting at 01000 wrongly rejects.

Franklin, 27 times

Franklin tops the duplication table at 27 jurisdictions. Washington's 26 hide the dataset's best joke: the District files 271 codes under the name, 25 states have one too, and Washington state has none at all. Meanwhile 78% of the 18,573 distinct place names exist in exactly one state — so name collisions are concentrated in a short head of founder-era names, and any importer that resolves a bare city name to a state works fine right up until it meets one of these.

The first digit is a map

ZIP codes ascend roughly east to west — 0 starts in New England, 9 ends in the Pacific ^[pub28]. The digit-8 Mountain West block is the thinnest at 2,794 codes, and digit-7 the fattest at 4,715, mostly on the strength of Texas. This is also why 25 of the 51 jurisdictions can be validated as a single contiguous prefix range while the other 26, Illinois and Georgia among them, need a set.

Reproducing these numbers

Everything above comes from one pass over the GeoNames US postal file ^[geonames], filtered to the 50 states and DC, grouping on the code, prefix, state, county and place-name columns. The per-state breakdowns live on the individual state pages, each with its own chart; the aggregation method is documented on the methodology page, and the same data is queryable through the free API. County definitions follow the Census Bureau's state geography guides ^[census-geo]. If you cite a figure, cite the denominator with it — and if you rerun the computation on a newer GeoNames snapshot and get a slightly different count, that is the dataset moving, not the method.