Random Names for Fiction and Characters

The most common naming failure in a first draft is not an implausible name, it is two names that occupy the same space in a reader's memory. Marcus and Marco, Kate and Clare, Sanders and Saunders: each is fine alone, and together they force the reader to re-read a line to work out who is speaking. A generator makes this worse than a phone book would, because a batch drawn from one pool shares its sounds.

The mechanical fixes are worth applying before you fall in love with anything. Give the named cast distinct first initials wherever you can. Vary the syllable count, so a two-beat name sits next to a four-beat one. Vary the ending sound, because names that rhyme blur faster than names that merely start alike. And say every name out loud with the dialogue tag attached — a name that is unpronounceable is a name a reader silently skips, which costs you the character.

Then check the whole cast together rather than one at a time. Lay the names out in a list, and look at what they say collectively: a cast whose every surname comes from one tradition is a decision, whether or not you meant to make it, and readers will read it as one.

The pools this page draws from are contemporary. They are not filtered by decade, which means a raw draw is right for a story set now and unreliable for one set in 1890 or 2140. For period work, use the generator for surnames — which move much more slowly than given names — and pick the given names deliberately, using the popularity note on each name page to check whether a name reads as of its era.

Region matters as much as period. Switching the country changes the pools rather than translating them, so a Turkish character gets Turkish names and a Japanese one gets Japanese names in native script. Two caveats: names are assembled given-name-first even where the convention is family-name-first, so write the order yourself from the separate fields; and the pools reflect present-day naming, including names that arrived recently through migration, which is realistic for a contemporary setting and anachronistic for a historical one.

For a secondary world, the honest answer is that this tool gives you a starting layer and not the whole cloth. Real names carry real history, which is why an unmodified real-world name in an invented world can feel like a costume. Writers commonly draw a real name, then bend the spelling or swap a syllable — which is a legitimate use of a generator, as long as the result stops being a specific real person's name rather than starting to be one.

Once a name survives the shortlist, spend two minutes checking it against the world outside your manuscript. Search the full name in quotation marks. If the first page of results is a living person with a public profile — a politician, an athlete, a local business owner — take another name, particularly if your character does anything unflattering. This is not a legal opinion, and nothing here is legal advice; it is the ordinary practice that publishers and studios formalise as name clearance, and it exists because the risk it manages is real.

The same check is worth running on the character's employer, their street and their phone number, all of which are invented in fiction and all of which occasionally turn out to belong to somebody. That is precisely why film and television use reserved ranges — the North American numbering plan keeps 555-0100 through 555-0199 unassigned for fiction, which is where every phone number this site generates for the United States comes from, and it is why a number quoted in a novel should come from that block too.

None of this is a reason to be timid. Common names are common, and a character named Sarah Jones does not defame the many real Sarah Joneses. The check exists for the specific case: an unusual full name, attached to a specific place or occupation, matching one identifiable living person.

The generator is a seeded, deterministic function rather than a list being shuffled: a seed string is hashed, the hash seeds an isolated random generator, and given name, middle name and surname are drawn in a fixed order from that country's pools. Given names and surnames are drawn independently, so the combination is a product of the draw and not a copy of any real person's record.

The pools are Faker's locale datasets — lists of names that are genuinely used in each language, which is why they read as plausible rather than invented. For the countries whose data has no middle names, the generator now leaves that field empty instead of borrowing an English one, and for Russian and Ukrainian, which ship genuine patronymics, the middle name is kept and written in Cyrillic like the rest of the name.

The name pages are a separate, hand-researched dataset: origin, meaning, variants and a popularity note per name, drawn from widely documented etymology rather than generated. They exist because the question a writer actually has is rarely "give me a name" and usually "does this name mean something I would regret".

Nothing you generate is stored, and nothing is registered to you. The same seed through the public API returns the same names, which is useful if you want a series bible whose cast can be regenerated exactly.