Most Common Last Names in the US: Top 100 From the 2020 Census

In April 2026 the Census Bureau released the last names reported in the 2020 Census ^{[release] [data-2020]}. It is the first new count in ten years, and most lists online still use the 2010 file. This page ranks the 2020 names and compares them with the 2010 and 2000 files ^{[data-2010] [data-2000]}. Every number comes from the Bureau's own files, and you can download the results as CSV.

The short answer is Smith, as it has been in every Census list since 1790 ^[brief]. The longer answer is a list that is slowly changing. Hispanic names entered the top 15 between 1990 and 2010, and Asian names have grown fastest since. If you need surnames for a form, a story or a test file, the last name generator gives you a random list in one click.

The 100 most common last names in the US

The table counts everyone with a recorded last name, of any age. The rate per 100,000 is the fairest way to compare names across years, because the number of recorded names changes. The 2010 column shows the rank each name held ten years earlier.

Eleven names belong to more than a million people each, from Smith down to Hernandez. The core of the list is old: eight of today's top 15 were also in the top 15 of the first census, in 1790 ^[release]. Predominantly Hispanic names now hold 26 of the top 100 places, and four predominantly Asian names appear: Nguyen, Patel, Kim and Chen. The full CSV goes down to number 1,000.

What is the most common last name in the US?

Smith, by a wide margin. The 2020 Census counted 2,369,644 people named Smith, which is 792.8 per 100,000. Put another way, about 1 person in 126 is a Smith. Johnson is second with 1,858,234, so Smith leads by more than half a million people.

The name is also one of the most widely shared across groups. Of the people named Smith, 68% are non-Hispanic White and 22.6% are non-Hispanic Black. Another 4.4% report two or more races and 3.5% are Hispanic. Smith even shows up on street signs: it is the 49th most common US street name, one place ahead of Johnson.

Smith's count fell slightly, from 2,442,977 in 2010. That does not mean fewer Smiths. In 2020 the Census recorded a last name for 90.2% of the people it counted ^[brief], against about 95.5% in 2010. When fewer names are recorded, almost every raw count falls. The rate per 100,000 is the better measure, and Smith's rate fell 4.3%.

What changed since the 2010 Census

The top five have not moved since 2000. Below them, the big change of the last 30 years is the rise of Hispanic last names. None of the 1990 top 15 was Hispanic; four were by 2000 and six by 2010 ^[brief]. Garcia, Rodriguez, Martinez, Hernandez, Lopez and Gonzalez all sit in the 2020 top 13.

Most of that climb happened between 2000 and 2010. Lopez went from 21st to 12th and Gonzalez from 23rd to 13th in that decade. Between 2010 and 2020 these names held their places. The one new step was Rodriguez passing Davis for eighth.

Since 2010 the fastest growth has come from Asian last names. The Census Bureau found that all but one of the fastest-growing names in the top 1,000 were predominantly Asian ^[release]. Our share-based comparison agrees. The table below ranks the names in the top 1,000 in both years by their change in rate per 100,000.

Higher up the list, Nguyen rose from 38th to 29th and Patel from 95th to 63rd. Chen entered the top 100 at 93rd. The biggest falls in the top 1,000 were small, about 8% in share, and all hit names held mostly by White Americans, such as Garrett, Thornton and Chandler. The changes CSV ranks all 972 names that were in the top 1,000 both times.

Two of those moves need a caution. Bell was 67th in 2000, fell to 106th in 2010 and came back to 76th. Hall did the same: 30th, then 45th, then 33rd. Their 2000 and 2020 ranks agree, so the 2010 figures look like the odd ones out. The Census files do not say why. If a list online puts Hall in the 40s, it was built from the 2010 file.

Why Doe is now the 78th most common last name

The strangest line in the 2020 file is Doe. In 2010, 7,066 people had the last name Doe, which ranked it 4,972nd. In 2020 the count is 262,774, and Doe sits 78th, between Peterson and Ward.

The Census Bureau explains this in its brief ^[brief]. Its staff deleted many entries that used John Doe or Jane Doe as a placeholder, such as 'JOHN DOE #2' and 'BABY DOE'. They kept plain 'John Doe' and 'Jane Doe', because those can be real names. As a result, the Bureau says, the Doe count is likely inflated.

So the real number of people named Doe is unknown, but the 2010 figure suggests thousands, not hundreds of thousands. It is a neat example of a placeholder leaking into official statistics. Our history of John Doe and other placeholder names explains how the name became the default stand-in.

Most common last names by race and Hispanic origin

The 2020 file splits each name's count into six groups. These are the Census Bureau's categories: Hispanic people of any race, and five non-Hispanic race groups. Ranking names within each group answers a different question from the national list. The Bureau's brief lists the names with the highest share in each group ^[brief]; the table below lists the most common names instead.

Most common Hispanic last names

Garcia leads, with 1,048,319 Hispanic people, followed by Rodriguez (1,010,137), Martinez (958,125) and Hernandez (953,863). These names are also among the most concentrated: over 91% of the people with each one are Hispanic. Hispanic last names cluster tightly. Just 35 names cover a quarter of the Hispanic population, against 258 for the country as a whole ^[brief].

Most common Black and African American last names

Williams is the most common last name among Black Americans, with 721,683 people. Johnson (623,699) and Smith (535,967) follow. Jackson is the most distinctive name in the top ten: 51.2% of the people named Jackson are Black. According to the Census brief, Washington has an even higher share, at 83.6% ^[brief].

Most common Asian last names in the US

Nguyen leads by a wide margin, with 507,213 Asian and Pacific Islander people. Lee (305,384), Patel (290,005), Kim (257,177) and Chen (223,126) follow. Lee is the one name in this top ten that is widely shared: only 43.9% of people named Lee are Asian, while 33.2% are White and 15.3% are Black.

Most common White last names

Among non-Hispanic White Americans, Smith (1,611,236), Johnson (1,055,206) and Miller (919,905) lead. Miller ranks seventh nationally but third here, because 81.5% of Millers are White. Anderson and Martin also rank higher in this group than in the national list.

Most common Native American last names

The American Indian and Alaska Native group is small in this file, about 1.8 million people. So common national names such as Smith and Johnson lead it. The two names right behind them are distinctive: Begay and Yazzie rank third and fourth, and over 92% of the people with each name are American Indian or Alaska Native. Locklear ranks eighth, at 66.1%.

How many last names cover the US population?

Americans share surnames less than you might think. After editing, the 2020 returns held about 7.8 million different last names ^[brief]. About 7.6 million of them, 98% of the total, belong to fewer than 100 people each. The public file lists the 156,621 names held by at least 100 people.

We also worked out the chance that two people picked at random share a last name. It is about 1 in 1,911. That sounds small, but it adds up quickly, like the birthday problem. In a room of 52 strangers drawn at random, the odds are better than even that two share a last name.

Real rooms beat those odds, because families and neighbours share names. The same maths explains why random name generators repeat sooner than people expect.

Using real last names in test data

These counts matter if you build sign-up forms, search or duplicate detection. A test file drawn evenly from a name list looks varied, but real customer data is not even. In a random sample of 10,000 Americans, you would expect about 79 Smiths, 62 Johnsons and 52 Williamses. Search, sorting and duplicate checks need to cope with that.

• Weight by frequency when you test matching. Use the per-100,000 column from the CSV as weights to build a realistic file with plenty of repeated surnames.
• Draw evenly when you need variety. A demo or a screenshot looks better without five Smiths, so an unweighted list such as our last name generator fits better.
• Restore the punctuation the Census removed. The files drop apostrophes, hyphens and spaces, so O'Brien is OBRIEN and De La Cruz is DELACRUZ. The letter Ñ becomes N, so Peña is PENA ^[brief]. Real data keeps these, and they break software. See our name validation test cases.

Download the data

Four files hold everything on this page. The top 1,000 CSV has counts, rates, 2010 and 2000 ranks and the share in each group. The group CSV lists the 25 most common names within each group. The changes CSV ranks names by their change since 2010, and the summary file holds the totals. They are CC BY 4.0, so reuse them with a link back. The Census files they come from are public domain.

Method note: the script downloads the 2020 workbook and the 2010 and 2000 files from www2.census.gov, about 25 MB in total. It checks that each row's six group counts add up to its total; all 156,621 rows do. It reproduces the brief's published figures exactly, including the 4.7% top-ten share and the 258 names for a quarter of the population. Shares use 298,898,231 people with a recorded last name in 2020 and 294,980,126 in 2010. The 2010 coverage of about 95.5% divides the second figure by the 308,745,538 people the 2010 Census counted.