Every number on Nameclover comes straight from official government birth-name data. Nothing is estimated or invented. Here is exactly what we use and what it does and does not mean.
For the year and country you pick, we take how many babies were given that exact name and divide by the total in the same dataset that year. So "1 in 533" means roughly one baby in every 533 in that data shared the name.
The denominator is the dataset's own total, not a separate birth count, on purpose. The U.S. figures count people who were issued a Social Security number, which is close to but not the same as every birth, so we never claim "all babies born." Keeping the top and bottom of the fraction from the same source is what makes the proportion honest. Numbers are always scoped to one country and one year. There is no worldwide rarity number, because the datasets are not comparable that way.
The count and the ratio always share one population: all records in that country and year, both sexes. When a result says "614 Dorothys among 3.6 million names in the 1987 records," 614 is every registration of that spelling that year, the pool is every name registered that year, and the ratio divides one by the other. Where a name is overwhelmingly one sex, we say so as a separate detail ("nearly all of them girls") rather than change the arithmetic. Pool sizes are rounded in the copy for readability; the exact totals live in the data files and this page's sources.
The full, sober version of this section lives at how the living estimate is made.
"Roughly 214,000 Dorothys are alive in the US today" is an estimate, not a count, and we always say so. Here is exactly how it is made: for every birth year, we take the recorded count of the name (separately for girls and boys) and multiply it by the probability that a person born that year is still alive at today's age, then sum across all years. Survival probabilities come from the SSA Actuarial Life Table (the 2023 period life table used in the 2026 Trustees Report; the lx column by sex, ages 0–119: survival from birth to age a is lx(a)/100,000). The "about 1 in N Americans" line divides the U.S. Census Bureau's Vintage 2025 national population estimate (341.8 million) by the living estimate.
Known gaps, so you can judge it: the method uses a period life table, which applies 2023 survival rates across whole lifetimes and therefore overstates survivors from early-1900s birth cohorts; it cannot see immigration (the name data counts US birth registrations only, so people who arrived with their name are missing); and pre-1937 births are under-registered in the source data. Displayed values are rounded because the precision is not real; the unrounded figures and the life-table file live in the repository. Estimates refresh with each annual data update. England & Wales has no living estimate yet; that waits for their life tables.
Every US result footer quantifies this instead of waving at it: we divide our year total by the official NCHS count of live births for the same year (NCHS births series, 1909–2018, extended through 2023 from the annual Births: Final Data reports). The story comes in three eras:
The rarity math deliberately does not correct for this. "1 in N" divides a name's count by the total of recorded names in the same dataset and year: numerator and denominator come from one universe, which is what makes the proportion defensible. Adjusting toward estimated births would mean inventing corrections we can't verify, so we quantify the gap and leave the arithmetic alone.
To protect privacy, the agencies hide very small counts. The SSA omits any U.S. name used fewer than 5 times in a year; the ONS redacts England & Wales counts under 3. When a name was clearly in use but is missing for one year, that almost always means it slipped under this line, which is exactly what makes it genuinely rare. We say "fewer than 5" (or 3) rather than guess a number.
The data records the exact spelling registered at birth, so Sofia, Sophia and Sofia-with-an-accent are counted as different names. We keep them separate and show related spellings as "related forms" rather than silently merging them, because deciding which spellings are "the same" is a subjective call we would rather not make for you.
Tiers (Extraordinary, Rare gem, Uncommon, Familiar, Popular) are based purely on the "1 in N" figure for that country and year. They are a friendly label on top of the real number, never a replacement for it.
Name pages add meaning, history and notable bearers on top of the rarity number. That editorial content follows a fixed sourcing standard, and every page is checked against it before it can publish.
A build-time check blocks any name page that fails these rules: an unverified or unpinned citation, a meaning with no confidence label, an etymology with fewer than two sources, a rarity caveat that does not match this page's wording, a figure phrased as a count of living people, or any placeholder number. Each page also shows two dates: when it was last verified by a person, and when the underlying data was last refreshed.
The underlying data and our processing are reproducible from the sources above. If a fact or a source looks wrong, it matters; the prepublish check exists precisely to catch claims the data does not support.
← Check a name