How SSA Baby Name Data Works: Sources, Limits, and What It Actually Measures

The data comes from Social Security card applications (Form SS-5), not from birth certificates directly. When a parent applies for a Social Security number for their newborn, the name on that application enters the dataset. Before 1937, coverage was spotty because Social Security did not exist until 1935 and most applications came from working-age adults. Coverage became essentially universal by the late 1940s.

Each record includes the child's first name, sex assigned at birth, state of birth, and year of birth. Middle names are not recorded. Hyphenated first names are recorded as a single entry (Mary-Jane, not Mary + Jane). Suffixes like Jr. or III are stripped.

Any name given to fewer than 5 babies of the same sex in a given year is excluded from the public file. This is a privacy measure, not a data quality filter. It means the published dataset undercounts rare names significantly. In any given year, roughly 4,000 to 6,000 unique name-sex combinations fall below this threshold and are suppressed.

The practical effect: if you are researching a truly unusual name, the SSA file may show zero occurrences even though 1-4 babies received it that year. The national file and the state-level files apply this threshold independently, so a name with 3 occurrences in California and 3 in Texas (6 total) might appear in the national file but not in either state file.

Two legislative events reshaped the dataset. First, Social Security expanded coverage in 1937, which dramatically increased the number of infant applications. Pre-1937 data is heavily biased toward names common among adults who applied for work-related Social Security numbers.

Second, the Tax Reform Act of 1986 required parents to list a child's SSN on tax returns to claim the dependent deduction. Before 1986, many parents delayed SSN applications for years. After 1986, applications became nearly universal at birth. This is why post-1986 data is considered the most reliable era for name frequency analysis.

Several important dimensions are missing:

• Gender-neutral tracking — the dataset assigns every name to exactly one sex category per record. A name like "Avery" appears in both the male and female files, but you cannot tell how many parents chose it specifically as a gender-neutral name. There is no non-binary or unspecified category in historical data.
• Race and ethnicity — SSA does not record race. You cannot determine from this dataset alone whether a name is more common in one ethnic group.
• Spelling intent — Kaitlyn, Katelyn, Caitlin, and Katelynn are four separate entries. There is no built-in way to group spelling variants.
• Pronunciation — two identically spelled names with different pronunciations (e.g., "Andrea" with stress on the first vs. second syllable) are merged.

Our name rankings pull directly from the most recent SSA national file and normalize for annual birth volume. When we say a name was "#47 in 2024," we mean it was the 47th most frequently applied-for name on SS-5 forms that year. We group common spelling variants in our explorer so you can see combined popularity, and we flag names that cross the gender boundary in our trend charts.

For historical trends, we use the full 1880-present file but add a reliability flag for pre-1937 data. Any trend line before 1937 should be interpreted with caution due to the adult-application bias described above.

SSA publishes the previous year's data each May, typically in the second or third week. The exact date is not announced in advance. The release includes both the national file and 51 state-level files (50 states plus DC). Territories like Puerto Rico and Guam are not included in the standard release. Each file is a simple CSV with three columns: name, sex, and count.