Last Name Generator
Generate random last names — surnames only, with no first name attached. Choose how many you need, then copy the list or download CSV and JSON. Free, no sign-up.
Updated
SchambergerKrajcikWatersAdamsHomenick
This last name generator returns surnames on their own. Pick a count of up to 100 and press Generate; copy a single name, copy the whole list one per line, or export CSV and JSON. Nothing is attached to the surname — no given name, no title, no profile — so the output drops straight into a surname column.
Writers use it to name a character when the first name is already settled, and to fill a cast list without every family sounding alike. Developers and QA teams use it to populate a last_name field with something more varied than “Test” repeated four hundred times, which is the fixture that hides real bugs: a surname column that only ever holds one short ASCII word will never reveal the layout that breaks on a hyphen or the query that breaks on an apostrophe.
The surnames come from Faker 9.9.0's locale data, the same pool behind the fake name generator. A generated surname is a real surname that real families carry — that is what makes it useful and what limits it. It identifies no particular person, and a batch can repeat.
For a first and last name together, use the fake name generator. For a surname inside a complete profile with an address and contact details, use the fake identity generator. For many rows with columns you choose, use the bulk test data generator.
Where surnames come from
| Type | Derived from | Examples |
|---|---|---|
| Patronymic | A parent's given name | Johnson, MacDonald, O'Brien |
| Occupational | A trade or office | Smith, Miller, Baker |
| Toponymic | A place or landscape feature | Hill, Ford, Washington |
| Descriptive | An appearance or character nickname | Young, Armstrong, Reid |
Most European-descended surnames fall into four families. The generator samples finished surnames; it does not construct them from these rules.
Are surnames gendered?
In English they are not: a surname is inherited whole and does not change form with the bearer's gender, which is why this generator has no gender selector. That is a property of English, not of surnames in general, and assuming it holds everywhere is a common source of broken software.
Several languages inflect the family name. Russian and Czech add a feminine ending — Ivanov and Ivanova, Novák and Nováková — so the same family carries two written forms. Icelandic mostly does not use inherited surnames at all: the name is patronymic and rebuilt each generation, giving Jónsson for a son and Jónsdóttir for a daughter. Spanish-speaking countries conventionally carry two surnames, the father's followed by the mother's. A schema with one fixed-form last_name field and a gender column that is expected to predict it will be wrong in all three cases.
Random surnames as test fixtures
A surname column is one of the most reliable places to find encoding and layout bugs, because real surnames contain the characters developers forget: the apostrophe in O'Brien, the hyphen in Smith-Jones, the space in Van der Berg, the accents in Müller and Nuñez, and the capital in the middle of McDonald. Sampling varied surnames into a fixture surfaces those before a user does.
This generator draws ordinary surnames from the configured locale, so it is a source of realistic values rather than a set of deliberate edge cases. When you need the hostile inputs on purpose — very long names, single characters, right-to-left scripts, names that are also SQL keywords — build them explicitly; the names that break software fixture pack collects them.
Surname frequency and what a random list is not
A random draw is not a population sample. The US Census Bureau publishes surname counts from the 2010 census — Smith, Johnson and Williams lead it, and roughly 6.3 million surnames appear in total, most of them held by fewer than a hundred people. Those published frequencies do not weight this generator, so a list of fifty names here will not match the distribution of fifty names drawn from a real population.
That matters when the output is meant to stand in for demographics. For a character or a test fixture it is irrelevant and often preferable, because a frequency-weighted list would return Smith repeatedly. For anything making a claim about how common a name is, use the Census file linked in Sources rather than this tool.
Frequently asked questions
Are these real last names?
Yes — they are drawn from lists of surnames that real families carry, not assembled from random letters. That is what makes them usable as test data and character names. A surname on its own identifies nobody, and the generator does not attach it to any real person or record.
Why is there no gender option for last names?
Because English surnames do not change form with gender — the same surname is carried by everyone in the family. Some languages do inflect: Russian and Czech add a feminine ending, and Icelandic rebuilds a patronymic each generation as -son or -dóttir. The English-language pool this tool uses has no such split, so a gender control would change nothing.
Can I generate a first name as well?
Use the fake name generator for a first and last name together, or the full identity generator for a surname inside a complete profile. This page deliberately returns the surname alone so it can be pasted into a single column.
Can the same surname appear twice in one batch?
Yes. Each name is drawn independently, so repeats are possible and get more likely as the count rises. Deduplicate after export if your fixture needs unique values.
Is the last name generator free?
Yes — free and unlimited, with no sign-up. Generate up to 100 surnames per batch, copy them and download CSV or JSON as often as you need.
Related pages
More generators
Free validation tools
Sources
- US Census Bureau — Frequently Occurring Surnames (2010 Census) — United States Census Bureau
- Faker — Localization and locale data coverage — Faker