Unisex names by country: how rare truly balanced names are

Truly unisex first names, used about equally for both sexes, are rare in every country we measured. In United States birth records, 645 of the 40,192 names given to at least 100 babies fall between 40% and 60% female: 1.6% of names, carrying 0.68% of all births. In Spain, Poland and Norway such names cover about 0.05% of people or less. And a name that is balanced in one country is usually firmly one gender in the next: Alexis is 45% female in Canada, 84% in the United States and almost entirely male in France.

The figures below come from official name statistics in ten countries, the same counted registers that feed our name-gender API. They were computed on 7 October 2026.

What counts as unisex here

A name is balanced in a country if at least 100 records carry it there and its female share is between 40% and 60%. "Records" are registered births in most countries and living residents in Spain, Poland and Switzerland, where that is what the statistics office publishes. The United Kingdom combines the three registers of England and Wales, Scotland and Northern Ireland.

This is a deliberately strict definition. Many names people call unisex, such as Jordan or Taylor, lean clearly one way in the counts. A name at 70% female is used for both sexes, but a reader of that name will still guess right seven times in ten. Balanced names are the ones where a guess is close to a coin toss.

Country by country

Two groups stand out. In the English-speaking countries, Switzerland and France, roughly one name in a hundred is balanced. In Spain, Poland, Norway and Austria the share of people with a balanced name is a tenth of that, and the balanced names that do appear are mostly from other languages: Jie, Yi and Xin in Spain, Tal, Chen and Bar in Poland, Thanh in Norway. Among names native to those countries, almost every one is clearly male or clearly female.

A unisex name in one country is rarely unisex in the next

The more useful finding for anyone inferring gender from names is how local "unisex" is. The same spelling can be balanced in one country, mostly female in another and mostly male in a third.

This is the same effect we document at scale in names that change gender by country: the gender of a name is a property of a name in a place.

Balanced worldwide is not the same as unisex

Pooling every country produces names that look balanced but are not balanced anywhere. Marian comes out at 53% female across all ten countries combined. In Poland it is a man's name, carried by more than 119,000 men and almost no women; in the United States it is 99% female. No one in either country would call it unisex. The worldwide figure is an average of two opposite answers.

For a name-gender lookup, this is the case that matters most. A worldwide probability near 50 can mean "genuinely used for both" or "used for one sex in each of two countries". Only the country tells them apart, which is why our API takes a country and why a result near 50 should be treated as unknown rather than rounded to the larger side.

Names balanced in several countries at once

103 names are balanced in at least two of the ten countries. The most widespread are:

• Quinn: balanced in the United States, Canada, the United Kingdom and Ireland, the clearest example of a name that is unisex across the English-speaking world.
• Harley (United States, Canada, Ireland, France), Sasha (France, Spain, Poland, Switzerland) and Casey (United States, Canada, Switzerland).
• Names from naming traditions where given names are shared by both sexes. Sikh given names are balanced in several countries: Amandeep in the United Kingdom, Canada, the United States and Spain, Mandeep in the United Kingdom, Canada and Spain; Tibetan Tenzin in France, Canada, the United States and Switzerland. Romanised Chinese names are balanced in four or five countries each (Yu, Yi), and Vietnamese Anh in three.

The romanised Chinese names are a measurement effect as much as a cultural one. Many Chinese characters with different gender associations share one romanisation, so Yu in a Western register merges several names that would be distinct in Chinese script. Chinese name gender explains why the original characters carry far more information.

Limits of these numbers

• Counts span decades. Most registers sum births over long periods, from 1880 in the United States. A name that moved from mostly male to mostly female over the century can look balanced in the total while being one-sided in any single year.
• Births and residents differ. Spain, Poland and Switzerland count living people; the others count births. The two are not directly comparable, especially for names that were popular long ago.
• Spellings are counted separately. Yael and Yaël in France, or Jaime and Jamie in the United States, are separate rows.
• Israel is not included. Israeli records are in Hebrew script and cannot be matched against the other countries' spellings.
• The threshold matters. At least 100 records excludes rare names, many of which are new and more often unisex. Lowering it raises the counts for every country.

What this means if you infer gender from names

Balanced names are a small share of any list, but they are not spread evenly: they cluster in particular countries and in communities with shared-name traditions. Three practical rules follow.

1. Pass the country. It is the only thing that separates Marian in Warsaw from Marian in Ohio.
2. Treat a probability near 50 as unknown. Rounding it to the larger side turns a coin toss into a wrong answer for nearly half the people with that name.
3. Check where your unknowns come from. If they cluster in one market or community, an aggregate result can be biased even when the overall unknown rate is small. The guidelines for gender inference in research show how to bound that effect.

Data

The counts come from the national statistics offices listed on the gender database page. The figures on this page were produced by scripts/unisex-by-country.php on 7 October 2026. Every name that appears in at least two countries, with its female and male counts per country, is in our open dataset of names that change gender by country (CC BY 4.0), which you can download and filter for balanced names yourself.