The Most Common Surnames in 58 Countries — and Why the Top 10 Means Something Completely Different in Vietnam and Italy

If you ask "what are the ten most common surnames in this country?", you will always get an answer. What almost nobody tells you is that the answer is worth wildly different amounts depending on where you ask.

In Vietnam, the top ten surnames cover roughly 71-76% of the entire population. In Italy, the top ten cover about 2-3%. That is not a rounding difference or a data-quality difference — it is a 25-fold gap in what the question even means. Knowing Vietnam's top ten tells you the surname of three people out of four. Knowing Italy's top ten tells you almost nothing about the next Italian you meet.

This article maps that spectrum across 64 locales covering 59 countries, using registry data rather than folklore. Along the way it kills one of the most-cited surname statistics on the internet.

The one number that actually matters: top-10 concentration

Surname distributions are power laws, but the steepness of the curve is a property of a country's history, not of surnames in general. Three forces set it:

• How long the country has had hereditary surnames. Turkey only legislated them in 1934 (the Soyadı Kanunu), and its surnames are descriptive and diverse. Vietnam's cluster of clan names is a thousand years older and far narrower.
• Whether surnames were assigned or grew organically. The Philippines got its surnames handed out from a printed list in 1849 (the Clavería decree) — which is exactly why dela Cruz sits at #1.
• Patronymics. Denmark, Norway, Sweden and Iceland froze "son of X" into hereditary names at a moment when only a handful of first names were in circulation. The result is a whole country of Jensen, Hansen and Andersson.

Everything else — how large the country is, how many surnames exist in total — matters much less than you would guess.

Top-10 surname concentration, country by country

Dashes mean the underlying source publishes a ranking but not a per-surname share we were willing to quote. Guessing one would have been easy; it would also have been made up.

Taiwan is not China (the top of the list is a different list)

This is the single most common error in English-language surname content: treating "Chinese surnames" as one distribution. Taiwan's Ministry of the Interior publishes a name-by-name registry of all 2,731 surnames in the country, and it does not look like the mainland at all.

Taiwan's top ten, from the registry:

1. 陳 (Chen) — 11.2%
2. 林 (Lin) — 8.3%
3. 黃 (Huang)
4. 張 (Zhang)
5. 李 (Li)
6. 王 (Wang) — 4.1%
7. 吳 (Wu)
8. 劉 (Liu)
9. 蔡 (Cai)
10. 楊 (Yang)

王 is the #1 surname of mainland China. In Taiwan it is #6, with less than 4.1% — while 陳, which is only #5 on the mainland, dominates at 11.2%. The whole island is nearly 53% concentrated in ten surnames, versus 42.9% on the mainland. Two Chinese-speaking populations, two genuinely different answers.

China's plateau: 王 is 7.25%, 李 is 7.19%

Here is the mainland's top ten, with shares from the 2007 Ministry of Public Security study:

Look at the top three. The gap between #1 and #2 is 0.06 percentage points — across 1.4 billion people, that is inside the noise of any census. "The most common surname in China" is not a question with a meaningful answer; 王, 李 and 张 are a plateau, not a podium. Every article that confidently crowns one of them is reporting a coin flip as a fact.

The plateau is easier to see than to describe. The first three bars are almost the same length; then the list falls away. That flat start is the whole point — there is no single leader, only a shared top.

The same plateau effect appears further down every list, in every country. In the United States, Davis, Rodriguez and Martinez sit at 395, 391 and 385 per 100,000 — the difference between rank #8 and rank #10 is statistical dust.

Nguyễn is 31%, not 38% — a zombie statistic

Search for Vietnam's most common surname and you will be told, essentially everywhere, that Nguyễn is 38% of the population. Wikipedia says it. Press articles say it. Software libraries hard-code it.

It traces back to a single 1992 sample by Lê Trung Hoa of 1,941 people.

That number is not repeated because it is good. It is repeated because it is the only one that was ever published in a citable form. Two large modern samples with opposite regional biases converge independently:

• VNTH01 (n = 1,682,729) → Nguyễn = 30.5%
• SG01 (n = 241,000, Ho Chi Minh City) → Nguyễn = 31.5%

Two datasets skewed in different directions landing within one point of each other is about as decisive as this kind of evidence gets. Nguyễn is ≈31%.

The companion claim, "Vietnam's top ten = 85%", comes from the same 1,941-person sample, and it has a second problem: it merges regional variants. Hoàng/Huỳnh and Vũ/Võ are northern and southern forms of the same surnames. Count them merged and the top ten reaches ~76% (SG01's own figure). Count them as separate written surnames — which is what a passport shows — and it is ~71%. The 85% figure is not just wrong; it is about a different object.

The general lesson travels well beyond Vietnam: when a statistic is cited by everyone but sourced to one tiny study, go find the big sample before you build anything on it.

Portugal and Spain: two surnames per person break the arithmetic

A Portuguese citizen carries two surnames — maternal plus paternal. So do Spaniards. This quietly makes two sentences both true and mutually contradictory-sounding:

• "45.7% of Portuguese people have one of the top ten surnames."
• "The top ten surnames account for 22.8% of all surname slots in Portugal."

Both are correct. The first counts people; the second counts positions. Because every person occupies two slots, all surname shares sum to ~200% rather than 100%, and the bearer figure is roughly double the slot figure. Silva is 9.4% of Portuguese people — but only ~4.7% of Portuguese surname slots.

Spain shows the same split: 34.4% of Spaniards carry a top-ten surname, but the top ten fill 17.55% of slots. García alone appears 2,915,761 times in Spain's national registry (INE, Censo 2025).

Which number you want depends entirely on the question. "How likely is a random Spaniard to be a García?" — use bearers. "If I read one surname off a form, how likely is it to be García?" — use slots. Mixing them produces confident nonsense, and a lot of published surname content mixes them.

Korea: the top ten depends on which alphabet you count in

South Korea's top ten covers 65.8% of the population — 김 (Kim) 21.5%, 이 (Lee) 14.7%, 박 (Park) 8.4%, then 정, 최, 조, 강, 장, 윤, 임.

But that 65.8% is the hangul figure. Count in hanja — the Chinese characters behind the names — and the top ten is 63.9%. The gap is real, not a data error.

The reason is that hangul is phonetic and hanja is not. 정 is written identically for 鄭 and 丁 — two unrelated clans, two different surnames, one hangul spelling. Same for several others. So "정" as a hangul string is more common than either of the hanja surnames inside it, and the merged list reshuffles: by hangul, 정 edges past 최, which is not the order you get from the hanja tables.

Neither number is wrong. They answer different questions: hanja counts lineages, hangul counts what is printed on the ID card.

The flat end of the spectrum: Italy, France, Ukraine

At the other extreme, the top-ten question nearly dissolves.

Italy is the flattest in Europe: Rossi is #1 at roughly 0.3% of the population, and the top ten together are ~2-3%. France is even lower at 1.71% for the entire top ten (INSEE). Martin is France's most common surname the way the tallest hill in a flat field is a hill.

The United States looks concentrated compared to Italy but is not: Smith is only 0.83%, and the top ten reach 4.90%. Behind that thin head sits a tail of roughly six million rare surnames.

Ukraine makes the point most vividly. Its top two are effectively tied:

• Мельник (Melnyk) — 107,878 bearers
• Шевченко (Shevchenko) — 106,340 bearers

A gap of 1,538 people across a country of tens of millions. And the tail is enormous: Ukraine has 707,685 distinct surnames, averaging about 64 bearers each. In a country like that, "rare and strange-sounding" is the statistical norm, not an anomaly — which is exactly why a Ukrainian surname list built from productive word-formation patterns (-енко, -ук, -ський) can look plausible and still be missing Мельник, the country's #1. Ours was, until we checked it against a registry.

The trap: share of the population vs. share of the list

This is the distinction that catches almost everyone, including our own tooling.

The top ten US surnames are 4.90% of Americans. But our curated list of 2,064 US surnames covers only about 45.36% of Americans — the rest live in that six-million-name tail, which no practical list contains. So inside that 2,064-name file, the top ten represent 4.90 / 45.36 = 10.80%. Both numbers describe the same reality. They just have different denominators.

If you ever compare "top-10 share in my dataset" against "top-10 share of the population" and find a discrepancy, the discrepancy is probably arithmetic, not error. The check that matters is whether the two reconcile through coverage. For Taiwan, ours does: 52.81% (in-list) × 99.98% (claimed coverage) = 52.79% of the population, matching the registry's 52.79% exactly. ⚠ Do not read that exactness as accuracy — both sides are computed from the same weights, so the identity is guaranteed to close whatever the file contains. It catches arithmetic errors, not wrong data. (The earlier 401-entry build read 48.14% × 109.8% = 52.84% against 52.81%.)

Vietnam is the exception that proves the rule: just 110 surnames cover 98.73% of Vietnamese, so there the list share and the population share are nearly the same number. That is a fact about Vietnam, not about methodology.

So what is "the most common surname"?

It depends on a question most people never ask: how much of the country does the top of the list actually own?

• In Vietnam and Korea, the top ten is a description of the population. Knowing it is knowing most people's names.
• In China and Taiwan, the top ten is real but the ranking inside it is fragile, and the two are not interchangeable.
• In Portugal and Spain, you must say whether you mean people or surname slots before the number means anything.
• In Italy, France, the US and Ukraine, the top ten is a trivia answer. It describes a rounding error of the population.

The next time you read "the most common surname in X is Y", the useful follow-up is not "is that right?" It is "and how many people is that, actually?"

Methodology / data as of 2026-07-17

Top-ten lists in this article are read directly from the weighted surname corpora that back our name generator (64 locales, format value⇥weight, weights calibrated to real frequency curves), not from memory or secondary summaries. Population shares come from the primary sources below; where a source publishes a rank but no share, the table shows a dash.

Primary sources:

Caveats worth stating plainly:

• Sources are of different vintages (2007 for China, 2015 for Korea, 2025 for Spain). Surname distributions move slowly, but the numbers are not simultaneous.
• Band figures (~4-6%, ~15-25%) are estimates from national top-100 tables, not registry sums. They are honest about their own precision.
• The Taiwanese registry is the only source here that is complete — every surname, down to the 643 borne by exactly one person. Everywhere else, the tail is an estimate.