How Many Surnames Does a Country Have? Taiwan Has 1,785. Ukraine Has 707,685. Italy Doesn't Know.

Ask how many different surnames exist in a country and you expect an answer in the same ballpark everywhere. You don't get one. Taiwan's Ministry of the Interior counted 1,785 surnames for 23.4 million people. Forebears counts 707,685 in Ukraine. Italy — a country with excellent statistics on almost everything else — has no official figure at all, and the leading Italian onomastics scholar has written an article explaining why one is not coming.

That is a spread of roughly 400× between the two countries that actually publish a count, and a hole in the middle where Italy should be. This article explains where those numbers come from, why they differ by orders of magnitude, and — the part almost every "most common surnames" listicle skips — why the number you get depends entirely on the counting threshold nobody mentions.

The short answer

Two things in that table are doing all the work. First, the East Asia / Europe split is real and enormous. Second, four different rows are counting four different objects — and that is not a footnote, it's the main story.

Why Taiwan has 1,785 surnames and Ukraine has 707,685

The gap is not a data artefact. It reflects two genuinely different systems.

Closed repertoires: an inherited, finite set

Chinese, Korean and Vietnamese surnames come from an ancient, essentially closed inventory of clan names. You do not acquire one; you inherit one, and no mechanism creates new ones. The consequence is extreme concentration. Taiwan's top ten surnames — 陳, 林, 黃, 張, 李, 王, 吳, 劉, 蔡, 楊 — cover 52.79% of the population, and the top hundred cover 96.62%. In Korea, Kim alone is 21.5% of the country, and the top ten reach 63.9%.

Taiwan's report shows the shape precisely: of its 1,785 surnames, only 4 are borne by more than a million people, while 740 have fewer than ten bearers. So the repertoire isn't just small — it's a handful of giants plus a long tail of near-extinct clan names, with almost nothing in between.

Open repertoires: names that were still being made

Slavic, Germanic and Romance surnames were formed comparatively late and from productive raw material: nicknames (Nesterenko, Perebyinis), occupations (Melnyk "miller", Kovalenko "smith's son", Ferrari, Schmidt), places (Poliszczuk "man from Polissia", Lombardi), and patronymics (García, Rodríguez, Fernández — all "son of" names). Every village produced its own. Nothing standardised them, and spelling drift multiplied each root into variants.

That's why Ukraine averages 64 people per surname while Taiwan averages about 13,100 — a factor of roughly 200. It isn't that Ukraine is bigger; it's that Ukraine's surname system never closed.

Italy is the extreme case of the same mechanism. Caffarelli lists why: a national standard language arrived late, so each region fossilised its own lexical forms; notaries, priests and clerks each spelled by ear; and the oral-to-written transition generated families of near-identical names differing by one letter — Pensavalle, Percivalle, Persivalle, all descended from Parsifal. A single place name, Bisceglie, hardened into Biceglia, Bisceglia, Bisceglio, Viscecchia, Visceglia, Visciglia, Viscillo, De Bisceglia, Di Bisceglia, Dibisceglie. Are those ten surnames or one? Your answer changes the national total by a lot — which brings us to the real problem.

Why the number depends on the threshold

Every published surname count is really the answer to the question "how many surnames have at least N bearers?" — and N is almost never stated in the headline. Change N and the number moves by an order of magnitude.

Spain. Roughly 892,000 distinct surnames were registered in 2024. But the list INE actually publishes is the "Lista de apellidos con frecuencia igual o mayor a 20 personas" — surnames with at least 20 bearers. Everything rarer is withheld for statistical confidentiality. So Spain has two perfectly true, wildly different answers depending on which document you read. (The "people per surname" figure for Spain also needs care: Spaniards carry two surnames, so surname positions are about twice the population, and the naive division understates the average.)

France. INSEE's Fichier des noms reports 218,982 different names nationally. Read the fine print: those are names attached to births between 1891 and 2000, with a minimum-frequency rule for release, and people born abroad excluded. It is not a count of surnames borne by living French residents. It's a count of surnames that appeared on French birth records across eleven decades. Different object, different number.

South Korea. The best illustration anywhere. The 2000 census found 286 surnames. The 2015 census found 5,582 — a twentyfold jump in fifteen years. Korea's naming tradition obviously did not change; the counting did. Of the 2015 total, 1,507 have hanja (Chinese characters) behind them and 4,075 do not — overwhelmingly naturalised citizens and multicultural families writing foreign names into Hangul, plus improved administrative-record methods. If you want a single fact proving that surname counts are definitions rather than discoveries, this is it.

The United States. The Census Bureau recorded about 6.3 million different surnames in 2010 — but 3.9 million of them (61.9%) were reported exactly once, a mix of genuinely unique names, spelling variants, and transcription errors from handwritten forms. The publicly released file uses a threshold of 100 or more occurrences and contains 162,253 names. Same country, same census: 6.3 million or 162,253, depending on N.

Taiwan. The reason Taiwan's 1,785 is unusually trustworthy is that it has no threshold at all. The Ministry of the Interior publishes surnames with a single bearer. There's no hidden tail. When a country with a household registry publishes without suppression, the count means what it says.

So: comparing 1,785 (no threshold) with 892,000 (no threshold, but only ~86,000-odd published) with 218,982 (≥30 births over a century) with 6.3 million (95.5% coverage, errors included) is comparing four different measurements. The 400× spread is real, but part of it is definitional, and honest writing has to say which part.

Italy: the country that cannot count its surnames

Italy is the best story here because its absence is principled, not lazy.

ISTAT does not collect surname repertoire data. The only serious counts Italy has ever had came from telephone directories. In the 1970s the linguist Emidio De Felice persuaded SIP, the state telephone monopoly, to hand over its subscriber lists on punched cards, and counted about 300,000 surnames — the basis of the first Italian surname dictionary (1978). In 1997 Jean-Pierre Pélissier repeated the exercise on the Italian phone book and got about 330,000. Today the conventional figure floats between 350,000 and 400,000, which would put Italy at roughly 170 people per surname — flatter even than Ukraine.

But those were counts of telephone subscribers: they undercounted women, under-30s, and small towns, and each listed head of household hid several unlisted relatives. And the channel is closed forever — mobile phones destroyed the universal directory.

Caffarelli's argument in Perché non possiamo contare i cognomi italiani is that modern databases don't rescue this. Any release would suppress counts below about 5, deleting exactly the enormous mass of one-bearer surnames that makes Italy interesting. The methodological questions have no agreed answer: are Diolosa and Diolosà one surname or two? (Barilla is just Barillà with the accent lost.) And the repertoire is a moving target — roughly half a million deaths a year retire surnames while naturalisations add new ones. Italy's honest answer is "we don't know, and here is why" — which is more rigorous than most countries' confident numbers.

What about given names?

Given names follow the same logic with the thresholds even less visible, and the registry data is thinner. Korea publishes surname statistics but not a comparable repertoire count for given names; Taiwan's report covers popular given names (家豪 and 淑芬 again topped the list) without a total. Forebears reports 70,414 distinct forenames for Ukraine at roughly 643 people per name — about a tenth as many distinct forms as surnames, which is the expected shape: given names are chosen from fashion, surnames are inherited from history. We'd rather flag that as a shape than dress up a number we can't source.

A note on name datasets — including ours

If you build or evaluate name corpora, there's a trap here worth naming explicitly, because it invalidates the whole exercise.

The size of a name file is not the number of surnames in a country. Our Ukrainian surname corpus has 2,562 entries against a national repertoire of 707,685 — a thin slice, by design. Our Vietnamese corpus has 298 entries and covers about 99.4% of the Vietnamese population, because the repertoire really is that concentrated. Our Taiwanese corpus has 2,707 entries and covers 99.98% of Taiwan. Ranking countries by file size would produce a ranking of our own work, not of the countries — Vietnam would look nearly 9× "poorer" than Ukraine when in fact its 298-name file is nearly complete and Ukraine's 2,562-name file barely scratches the surface.

The right metric for a corpus is population coverage, not entry count. Those are different denominators, and mixing them is the same class of error as comparing Taiwan's unthresholded 1,785 with America's thresholded 162,253.

Conclusion

There is no single answer to "how many surnames does a country have," and the honest range is enormous: from 1,785 in Taiwan to 707,685 in Ukraine to somewhere-around-350,000-but-nobody-really-knows in Italy. Most of that spread is real — closed clan repertoires versus open, still-generating ones. But a meaningful slice of it is definitional, and you can't tell which is which until you find the threshold. Korea's count grew twentyfold in fifteen years without a single new Korean surname being invented. Whenever you see a surname count without a threshold attached, that's the number to distrust first.

Methodology and sources

All figures verified 2026-07-17. Where a source is an aggregator rather than a state registry, it is labelled as such.

Known limitations. The Ukrainian figure is the weakest number in the table: it comes from an aggregator with an undisclosed threshold and an undisclosed population base, and it cannot be cross-checked against a registry, because none is published. It is included because it is the only figure available and because its order of magnitude is corroborated by the structure of Slavic surname formation — not because it carries registry authority. The Spanish "people per surname" column is distorted by Spain's two-surname convention and should be read as indicative only. "People per surname" is omitted for France because the underlying file counts births over 1891–2000, not living bearers.