Names of United States US

Up to five names at a time.

Countries whose names look alike

Measured as the overlap between each country's top 300 given-name forms, weighted by how common each is — the fuzzy-set approach of Lauc (2024), on a truncated inventory.

• CACanada71%
• GBUnited Kingdom59%
• AUAustralia51%
• ITItaly26%
• PHPhilippines23%

Official registers

• Census Bureau — Surname Frequency 20102010headline (surnames)

  US Census Bureau. Counts: residents. surname. 162,081 forms. 2010. Licence: Public Domain.

  ▶What we know about this register

  162,253 surnames occurring >= 100 times in the 2010 Census, with rank, count and the race/Hispanic-origin breakdown. NOT a full surname list: list_coverage was corrected from `full` to `partial` because the >=100 threshold truncates the tail hard. Per Comenetz, J. (2016), 'Frequently Occurring Surnames in the 2010 Census' (US Census Bureau), the full tabulation holds ~6,299,033 distinct surnames, so this file publishes 2.6% of them; the other ~6.14M are collapsed into a single 'ALL OTHER NAMES' row (rank 0, count 29,312,001 = 9.94% of the population, 9936.97 per 100k). That row is NOT a name and is excluded from the name rows — its magnitude is recorded here so the truncation stays visible. The six race/Hispanic-origin columns (pctwhite, pctblack, pctapi, pctaian, pct2prace, pcthispanic) plus prop100k/cum_prop100k have no home in the 12-column schema and are preserved as key=value pairs in data_note — this is what makes breakdown=[ethnicity] true rather than aspirational; the previous ingest declared it and dropped it. A cell of '(S)' means the Bureau suppressed that percentage for disclosure avoidance: suppressed cells are OMITTED from the pairs and named in a `suppressed=` list, so a suppressed value is never read as a zero. Two genuine surnames are NaN traps: 'NULL' (rank 4910, 7,170 bearers) and 'NAN' (rank 30277, 773 bearers). An earlier generic ingest read the CSV with default NaN handling and destroyed 'NULL', emitting a null-named row; it also produced only 10 of the 12 canonical columns (no evidence, no data_note), typed name_romanized/sex/region_code as INTEGER, and left 16 all-null junk rows. All fixed: the parser uses keep_default_na=False with na_values=[]. DOWNLOAD IS BLOCKED from this environment: www2.census.gov and www.census.gov sit behind Cloudflare and return HTTP 403 to plain curl, a browser User-Agent, a Referer, and every proxy configured in official/.env (whose credentials are in any case expired — 407 Auth failed). The parser therefore fetches byte-identical Wayback captures of the same census.gov URLs via the `id_` raw modifier; pass --url to go direct from a network that can reach census.gov. Parser scripts/parse_us_census_surnames.py; raw zip retained under raw/US/ (the previous ingest kept none). A 2020 edition DOES exist (published April-May 2026) and is registered as us-census-surnames-2020 — an earlier note here wrongly said none existed.

  Last verified 2026-07-30.

• Census Bureau — First Name Frequency 2020 by sex2020headline (given)

  US Census Bureau. Counts: residents. given. 51,296 forms. 2020. Licence: Public Domain.

  ▶What we know about this register

  THE FIRST OFFICIAL US GIVEN-NAME FREQUENCY LIST IN THIS REGISTRY: 53,615 distinct first names from the 2020 Census, emitted as 98,562 rows (51,474 F + 47,088 M) because the file carries real MALE and FEMALE counts and `sex` is both a schema column and part of the combined build's dedupe key — one row per sex makes sex queryable instead of burying it in data_note; the both-sexes figure is kept as total_both_sexes=. This is census-wide population data, categorically different from SSA birth-cohort baby-name files. 2020 release, published April-May 2026. SCHEMA BREAK from 2000/2010: plain .xlsx (real header on the third row) and the race/Hispanic columns are ABSOLUTE COUNTS, not percentages — they sum exactly to the frequency, which is the check that validates the parse. More tightly protected, not less: counts are NOISE-INFUSED (+/-3 at 95%, DRB approval CBDRB-FY26-058) and the universe is doubly gated to names with frequency >= 100 in BOTH 2010 and 2020. The noise is why published rows can fall below 100 (minimum observed 92), so `count` is no longer a clean threshold and min(count) must not be read as the cutoff. The residual 'ALL OTHER NAMES' row (18,794,706 people) is excluded. A companion Names2020_FirstNames_RaceHispanic.xlsx gives the same names by race instead of sex and is not ingested (sex is the more useful axis for this schema). census.gov returns HTTP 403 to scripted clients (Cloudflare); the parser fetches byte-identical Wayback captures of the same census.gov URLs. Parser scripts/parse_us_census_surnames.py.

  Last verified 2026-07-30.

• Census Bureau — Surname Frequency 20202020

  US Census Bureau. Counts: residents. surname. 156,450 forms. 2020. Licence: Public Domain.

  ▶What we know about this register

  156,619 surnames. 2020 release, published April-May 2026. SCHEMA BREAK from 2000/2010: plain .xlsx (real header on the third row) and the race/Hispanic columns are ABSOLUTE COUNTS, not percentages — they sum exactly to the frequency, which is the check that validates the parse. More tightly protected, not less: counts are NOISE-INFUSED (+/-3 at 95%, DRB approval CBDRB-FY26-058) and the universe is doubly gated to names with frequency >= 100 in BOTH 2010 and 2020. The noise is why published rows can fall below 100 (minimum observed 92), so `count` is no longer a clean threshold and min(count) must not be read as the cutoff. Six race/Hispanic-origin counts are preserved in data_note as n_white, n_black, n_aian, n_api, n_2prace, n_hispanic. The 'ALL OTHER NAMES' residual (36,257,637 people) is excluded as a non-name; per the Bureau the full 2020 tabulation has 7,765,578 distinct surnames (9.4M raw, 1.6M edited out), so this file publishes ~2%. NOTE the 2020 editing removed the surname 'NULL' that is present in both 2000 and 2010 — verified against the raw xlsx, so it is an upstream editorial change, not a NaN-parsing artifact; 'NAN' (rank 25954) survives and is genuine. 2020 also fixed the N-tilde mojibake of earlier editions by mapping to N (PENA). A companion _WithNegatives.xlsx holds the pre-optimization noisy counts (67,072 negative cells) and is deliberately NOT ingested. census.gov returns HTTP 403 to scripted clients (Cloudflare); the parser fetches byte-identical Wayback captures of the same census.gov URLs. Parser scripts/parse_us_census_surnames.py.

  Last verified 2026-07-30.

• Census Bureau — Surname Frequency 20002000

  US Census Bureau. Counts: residents. surname. 151,491 forms. 2000. Licence: Public Domain.

  ▶What we know about this register

  151,671 surnames occurring >= 100 times in the 2000 Census — the earlier edition of the same series as us-census-surnames-2010, giving a second time point on identical columns. Zip member is app_c.csv (the 2010 zip uses Names_2010Census.csv). Same >=100 threshold, hence list_coverage=partial, but note it carries NO 'ALL OTHER NAMES' residual row, so the below-threshold population cannot be quantified from this file the way it can for 2010. The six race/Hispanic-origin columns (pctwhite, pctblack, pctapi, pctaian, pct2prace, pcthispanic) plus prop100k/cum_prop100k have no home in the 12-column schema and are preserved as key=value pairs in data_note — this is what makes breakdown=[ethnicity] true rather than aspirational; the previous ingest declared it and dropped it. A cell of '(S)' means the Bureau suppressed that percentage for disclosure avoidance: suppressed cells are OMITTED from the pairs and named in a `suppressed=` list, so a suppressed value is never read as a zero. Same NaN traps, both genuine surnames here too: 'NULL' (rank 4590, 7,072) and 'NAN' (rank 44895, 450). DOWNLOAD IS BLOCKED from this environment: www2.census.gov and www.census.gov sit behind Cloudflare and return HTTP 403 to plain curl, a browser User-Agent, a Referer, and every proxy configured in official/.env (whose credentials are in any case expired — 407 Auth failed). The parser therefore fetches byte-identical Wayback captures of the same census.gov URLs via the `id_` raw modifier; pass --url to go direct from a network that can reach census.gov. Parser scripts/parse_us_census_surnames.py (--year 2000).

  Last verified 2026-07-30.

• SSA — National Baby Names (1880–present)2025

  Social Security Administration. Counts: newborns. given. 20,558 forms. 1880–2025. Licence: Public Domain.

  ▶What we know about this register

  names.zip = national; one txt per year: name,sex,count. Names <5 suppressed. SSA CDN requires Referer header — fetched via custom handler.

  Last verified 2026-05-09.