Data and methods

The data

Baby-name counts come from the US Social Security Administration (SSA), which tabulates the first names on Social Security card applications for babies born in the US. SSA publishes every name given to at least five babies of the same sex in a year: nationally since 1880 and by state of birth since 1910. The files are updated each May; this site uses the release with data for 2025, tabulated from records as of March 1, 2026. The files are public domain.

Limits worth knowing: many people born before 1937 never applied for a card, so early years are incomplete (in 1910 the records cover about a quarter of births; by the 1950s almost all). Spellings are counted separately and not merged by SSA. Hyphens and spaces are removed (Mary-Kate is Marykate). Names with fewer than five babies in a year, nationally or in a state, are left out, which hides rare names, especially in small states. SSA does not edit the data, so a few records have the wrong sex.

Ranks and counts

Ranks follow SSA's own order: by number of babies of that sex, with ties in alphabetical order. A name's "babies in 2025" is the official count for that sex. "Per million" or "per 1,000" figures divide a name's count by all babies of that sex in SSA's list for the year, which keeps years with different numbers of births comparable.

Counting every spelling ("true popularity")

We add up spellings that are pronounced the same in American English and used for the same sex, such as Sophia and Sofia, or Jackson, Jaxon and Jaxson. Candidate spellings are found with phonetic codes (Double Metaphone and NYSIIS) and our own vowel-aware sound key, then accepted only if they also pass these checks: the girls' share of babies differs by no more than 35 points between the spellings; the Carnegie Mellon Pronouncing Dictionary, where it lists both, gives the same pronunciation; and the first sound and number of syllables match. An editor's list of exceptions overrides the rules (for example Ava and Eva are kept apart). Every spelling must match the family's most common spelling directly, which stops chains like Ella to Ela to Eli.

Nicknames and forms that sound different are never added: Theo and Theodore, Sophie and Sophia, Isabel and Isabella stay separate and are shown as related. Each name page lists every spelling in its family with its own official count and rank, so you can check the total. Families with at least 100 babies in 2025 get a combined rank; the official rank is always shown too.

States

State maps use the latest five years (2021–2025) to smooth small numbers: babies given the name per 1,000 babies of that sex in the state's SSA list. "More common than nationally" compares that rate with the national rate. Lists of names that stand out in a state require at least 15 babies in the state that year (or 50 over five years) so that a handful of births cannot dominate. States with fewer than 20 recorded names for either sex in a year have no separate page for that year.

How many classmates share a name

We divide a name's babies by all registered births that year (from the National Center for Health Statistics, not just SSA's list) to get the share of babies with the name, then work out the chance that at least one other child in a class of 20 born the same year has it: 1 minus (1 minus the share) to the power of 19. A class of 20 sits between the average public elementary class sizes in federal school surveys (19.1 in 2020–21 and 21.2 in 2015–16), and a typical public elementary school has about 430 pupils. National figures use 2025; state figures use the latest year with state birth totals. Real classes mix birth years and vary by school, so treat these as rough odds.

Ages and people alive

For each year since 1900, we multiply the babies given a name by the share of people born that year who are still alive, from the life tables in SSA's 2026 Trustees Report (intermediate assumptions), then add them up. That estimates how many US-born people with the name are alive in mid-2026 and how old they are: the median age, the middle half, and the age profile chart. The method follows the approach FiveThirtyEight described in 2014. It counts only people born in the US (immigrants are missing), assumes people with every name live as long as average, and inherits the gaps in early records.

Styles

Style lists (vintage revival, truly unisex, timeless, new names, names that defined a decade, state favorites and signature names) are computed from each name's share of babies over time, with the exact rule printed on each list. Thresholds are kept the same every year so the lists can be compared.

Meanings, origins and pronunciation

Meanings and origins are written in our own words from at least two independent references, led by scholarly sources such as the Dictionary of Medieval Names from European Sources and, for biblical names, the text with a scholarly translation's notes, alongside Wiktionary and Wikidata for the language path and related forms. Each record carries a certainty level: established, probable, uncertain, disputed or unknown. When an etymology is folk tradition rather than documented, we say so; modern names whose origin is not documented are labelled that way rather than given an invented meaning. Spelling variants and short forms point to the name they come from.

Pronunciations are the usual American English ones, written as a respelling with the stressed syllable in capitals, plus IPA, from Wiktionary, the Carnegie Mellon Pronouncing Dictionary and our editors. Audio clips are generated from that reviewed pronunciation with the open-source Kokoro text-to-speech model, so a name is only voiced when we have a pronunciation we trust.

Meaning themes ("names that mean light") and origins are tagged by an editor reading each record, not by matching words. A theme is tagged only when the documented roots carry it; folk etymologies, famous bearers and sound never count, and names of uncertain or disputed origin are not grouped. Each name has one origin: the language its meaning comes from, which may differ from the language it reached English through (Mateo is filed as Hebrew).

Similar, sibling and middle names

"Names with a similar history" compares each name's share of babies year by year since 1880 (smoothed over three years) with every other name's, and keeps the closest curves among names that were about as common, within a factor of eight, around the name's peak. Other spellings of the same name, near-spellings and likely clerical slips (a dropped or swapped letter in a name 100 times rarer than its neighbor) are left out. Sibling ideas use the same comparison across sexes and drop names with the same first letter or ending.

There is no official record of middle names, so middle-name ideas follow sound rules rather than popularity: a different number of syllables from the first name, a different first sound and ending, no run-on where the names meet (a name ending in a vowel next to one starting with a vowel, or the same consonant twice). Initials depend on the surname, so the full-name checker tests them instead. Candidates are names given to at least 2,500 babies of the sex in the last five years.

Nicknames are reviewed by an editor for each of the most used names, starting from a public dataset of name equivalents (carltonnorthern/nicknames, which was built for genealogy and contains many pairs that are not nicknames today), the name's etymology and dictionary entries. Only forms in real US use are kept, marked traditional or modern.

Which names have pages

13,370 names have a full page: those given to at least 1,000 babies in ten or more years, or to at least 500 since 2006. They cover the overwhelming majority of babies. Every other name SSA has recorded, about 106,000 in all, can be looked up with How common is my name?, without a page of its own, because a page with only a few scattered years of data would tell you little.