The Name Lab
Everything on this page is computed from one file: the Social Security Administration’s record of 105,924 first names, covering 1880 to 2025. Every name given to at least five babies of one sex in a year is in it. We measure each name as a share of that year’s recorded births, so 1912 and 2012 compare fairly. It’s the same dataset and method behind our 2026 forecast. Five findings, five charts.
The collapse of the top ten
In 1880, 44% of American boys were given one of the year’s ten most popular names, and the hundred most popular covered 80% of them. Girls were always named more inventively (their top ten took 25%), but for both sexes naming was closer to inheritance than invention. The decline took ninety years to get going and thirty to finish: across both sexes, the top-ten share fell from 26.7% in 1880 to 14.6% in 1970, then to 4.4% today.
The ten most popular boys’ names now reach 8% of boys, the girls’ ten just 7% of girls. Even the top hundred names of each sex, combined, cover less than a quarter of all babies (23.8%). Parents stopped naming children after each other. The rest of this page is about what they did instead.
Data table
| Year | Boys top-10 | Girls top-10 | All top-100 |
|---|---|---|---|
| 1880 | 44.2% | 24.6% | 65.9% |
| 1900 | 34.2% | 19.5% | 58.3% |
| 1920 | 31.6% | 23.2% | 55.3% |
| 1940 | 33.5% | 23.1% | 57.8% |
| 1960 | 29.1% | 17.0% | 54.1% |
| 1980 | 23.6% | 17.1% | 50.7% |
| 2000 | 13.4% | 10.2% | 36.7% |
| 2025 | 8.0% | 7.0% | 23.8% |
The -n takeover
Fashion in names is mostly fashion in sound. In 1950, 14% of boys got a name ending in -n, about what it had been in 1880 (15%). Then the ending went vertical: Brandon and Justin in the eighties, Jordan and Austin in the nineties, then the Aiden rhyme family. By 2011, 36% of all boys born in America carried a -n name (Mason, Logan, Ethan, Jackson, Jayden), more than the next five endings combined (33%).
The wave is receding, but only to 28%, still double the 1950 level. Girls run an older monoculture: -a, at 38% of girls last year. The old champion -e, which led the table in 1880 at 37%, has fallen to 18%.
Data table
| Year | -n | -r | -s | -o | -e | -l |
|---|---|---|---|---|---|---|
| 1880 | 15.3% | 6.7% | 16.7% | 0.8% | 12.2% | 4.6% |
| 1920 | 15.1% | 6.2% | 12.6% | 1.9% | 12.3% | 6.7% |
| 1960 | 15.3% | 3.1% | 10.3% | 1.3% | 8.4% | 10.5% |
| 1990 | 25.4% | 7.7% | 8.4% | 3.3% | 7.1% | 9.4% |
| 2011 | 36.4% | 8.8% | 6.4% | 4.2% | 6.8% | 7.0% |
| 2025 | 28.0% | 9.5% | 8.5% | 7.7% | 6.9% | 6.7% |
Killed by culture
A name can survive a century of fashion and not survive one news cycle. 4,249 girls were named Alexa in 2014, the year Amazon put the name on a speaker you shout at; 2025 saw 243, a 94% collapse in share. Katrina lost 93% of its share after the 2005 hurricane. Isis, an ordinary goddess name given to 401 girls as late as 2014, fell 65% once the acronym took over the news.
Two of the panels hide slower deaths. Hillary collapsed twice: from 3,690 girls in 1992 to 554 two years into its First-Lady era, then by half again after 2016. Elsa is the strangest. Frozen doubled it overnight (566 girls in 2013, 1,140 in 2014), then parents realized the name belonged to a cartoon; at 216 girls it now sits below its pre-movie level. Karen needs no retelling: down 47% since 2020, 175 girls last year.
Share of girls born each year, 1990–2025; each panel on its own scale. Hillary includes the Hilary spelling.
Data table
| Name | Event | Girls, event year | Girls, 2025 |
|---|---|---|---|
| Alexa | Amazon Echo launches, 2014 | 4,249 | 243 |
| Isis | ISIS takes Mosul, 2014 | 401 | 128 |
| Katrina | Hurricane Katrina, 2005 | 1,328 | 83 |
| Hillary | the election, 2016 | 199 | 91 |
| Elsa | Frozen premieres, 2013 | 566 | 216 |
| Karen | "Karen" meme peaks, 2020 | 332 | 175 |
The one-way door
Every name in this chart began as a boys’ name. Leslie became majority-girl in 1946, Ashley in 1964, Aubrey in 1974, Morgan in 1980, Madison in 1984 (the year Splash put it on a mermaid as a joke, because Madison wasn’t a name), and Taylor in 1990. None came back. Once a name reads as a girls’ name, parents of boys leave and don’t return; the door swings one way.
The blue line is the exception that proves the mechanism. Casey touched 50.1% girls in 1995, a coin flip, then retreated all the way to 18%. It’s the only name here that reached the line without crossing it for good, and the ambiguity seems to have cost it both audiences.
Data table
| Name | Crossed 50% | % girls, 2025 |
|---|---|---|
| Leslie | 1946 | 92% |
| Ashley | 1964 | 99% |
| Aubrey | 1974 | 97% |
| Madison | 1984 | 100% |
| Morgan | 1980 | 65% |
| Taylor | 1990 | 66% |
| Casey | never | 18% |
Names fade slower than they rise
We assumed the fad shape: slow climb, spike, crash. The data says the opposite. Of the 750 names that peaked at one baby in a thousand or better between 1900 and 2000, 468 completed a full arc in the record, from a tenth of their peak, up, and back down to a tenth. The median climb took 18 years; the median fall took 28. The average trajectory is lopsided the same way: ten years before its peak, a name sits at 44% of its maximum; ten years after, it still holds 54%.
Linda, which peaked in 1948, took 11 years up and 22 down; Jason (1974) took 7 up and 36 down. Names don’t crash; they deflate, one baby shower at a time. The panels above are the exceptions: fashion lets a name down gently. It takes a hurricane, a war, or a trillion-dollar company to kill one overnight.
Data table
| Years from peak | Average share of peak |
|---|---|
| -40 | 9.0% |
| -30 | 13.0% |
| -20 | 21.0% |
| -10 | 43.6% |
| 0 | 100.0% |
| +10 | 54.2% |
| +20 | 30.9% |
| +30 | 18.6% |
| +40 | 11.4% |
Methods & data
Source: the Social Security Administration’s national baby-name file, which counts babies by name, sex, and year of birth from Social Security card applications, 1880–2025. The SSA omits any name given to fewer than five babies of a sex in a year, so every series here has a floor and rare-name tails are truncated. All popularity figures are shares of recorded same-sex births in that year (combined across sexes where noted), which keeps eras comparable even though registration was far from universal before the 1930s. Analyses run over all 105,924 names in the file; nothing is sampled.