Two federal datasets, two different questions, and a large hole in the middle of both. Worth understanding before trusting any number on this site.
The Census Bureau publishes every last name reported by at least about 100 people. That file holds 156,621 names covering 262,640,594 people. The tabulation base is 298,898,231, so 36,257,637 people — 12.1% — have a last name too rare to publish. For first names the gap is smaller but still large: 18,794,706 people, 6.2%.
The withholding is deliberate and correct. A name held by three people in a county is identifying, and a census that published it would be a census people stopped answering honestly. But it means this site cannot tell you anything about roughly one American in eight, and the ones it cannot tell you about are exactly the ones most curious about the answer.
The floor is also softer than "100" suggests. Disclosure avoidance adds statistical noise before publication, so 617 last names appear with counts below 100 — the lowest is 92. Treat every count here as very close to right rather than exact, and small counts as approximate.
The rank and count on this site come from the 2020 Census: people alive and counted in 2020. The birth curve comes from Social Security card applications: babies born in each year since 1880. They are not two views of one number, and a name can look completely different in each.
MARY is the clearest example. It is the 7th most common first name among living Americans, held by more than 1.7 million people — and it was given to 2,230 babies in 2025, down from 74,171 at its peak in 1921. A name can be everywhere and nowhere at once, depending on which question you ask.
Neither dataset covers everyone. The Census counts residents; the SSA counts people who applied for a card, which since 1987 has been effectively all babies but was patchier before that. Early years in the birth data undercount badly. So do the most recent ones, because a card application is filed some time after a birth.
How many people share a full name. No public dataset records first and last names together. Multiplying the two frequencies assumes they vary independently, and they emphatically do not — the composition data shows names sorting strongly by ancestry, and first names sort the same way. The product would be badly wrong in both directions, so this site does not compute it.
Anything outside the United States. Many national statistics offices publish name data — the UK, the Netherlands, the Nordics, Spain, Australia and others — but coverage is concentrated in Europe and the Anglosphere, definitions differ, and most of the world's population is not represented at all. A single global ranking would be invented rather than measured.
Where a name comes from. The demographic columns count who reports a name now. That is a fact about American history and migration, not about etymology.