LuckPicker

Tool

Random Word Generator

Draw English words with length and starting-letter filters — and see the pool shrink.

Random Word Generator

234 of 234 words match. Every filter you add shrinks the pool, and a small pool means repeats.

Draw English words with a length range and an optional starting-letter filter, one at a time or up to twenty at once.

The surviving pool size is shown, and it is the number worth watching. A three-letter filter on a curated list can leave two words, and drawing from two words is not really drawing.

Words are drawn without replacement within a batch, so a request for ten gives ten different words rather than ten independent draws that might repeat.

How the Random Word Generator works — and why it's fair

Two filters run before the draw: a length band and an optional starting substring. Both are applied as a conjunction, so a word must satisfy every active condition, and what remains is the sample space. The draw over that survivor set is an unbiased uniform pick, using rejection sampling so no position in the filtered list is favoured.

Multi-word requests use a partial Fisher-Yates draw rather than repeated independent picks. That guarantees the batch contains distinct words, which is almost always what people want — a batch of ten containing the same word twice looks like a fault even when it is a correct independent draw.

The filters compound faster than they appear to. A length band of five to six leaves 153 of the 234 words; adding a starting substring of 'qu' leaves five. The tool prints the count precisely so that collapse is visible before you draw rather than inferred from repeated results.

The corpus is 234 words, deliberately curated rather than exhaustive, and the composition matters for what the draws feel like: it is weighted toward concrete nouns, because those are what word games, writing prompts and vocabulary drills actually need. A dictionary-complete list would be mostly inflections, technical terms and words nobody would use in a game.

That curation is also a real limitation, and the honest version of it is that the tool cannot draw a word that is not on the list. There is no long tail here. If you need genuine dictionary coverage, this is the wrong tool.

Every word is equally likely within the surviving pool, with no weighting by frequency or familiarity. A rarely-used word has exactly the same chance as a common one, which is different from how a Scattergories draw should work and is why that is a separate tool.

When the Random Word Generator is fair — and when it is not

What it does guarantee

  • Every word passing your filters has exactly the same chance, with no frequency weighting.
  • A multi-word batch is drawn without replacement, so it contains distinct words.
  • The surviving pool count is displayed before the draw, so a narrow filter is visible rather than inferred.

What it does not

  • The corpus is curated and finite. Words not on the list have zero chance, and there is no long tail.
  • It is weighted toward concrete nouns by composition, so the draws are not representative of English.
  • Filters compound. A length band plus a starting letter can leave a pool of two, which the count will show.

Two worked examples

Length 5 to 6, no letter filter

  • 153 of the 234 curated words survive — most of the corpus sits in that band.
  • Each has about a 0.65% chance on a single draw.
  • A batch of 10 draws 10 distinct words from that pool.

Length 5 to 7, starting with 'qu'

  • The pool collapses to five — quarry, quartz, quiver, quilt, quince.
  • Each has a 20% chance, and a batch of 10 returns only the 5 that exist.
  • The count is printed before you draw, so the collapse is visible in advance.

What people draw words for

Writing exercises are the commonest use, and the constraint is usually the point: a prompt to write a paragraph containing three drawn words works because the words were not chosen, which forces connections a writer would not have made.

Vocabulary and spelling practice uses the length filter directly, since a spelling list needs words of comparable difficulty rather than a random spread from three letters to twelve.

Improvisation and word games use it as a neutral source — a word nobody in the room selected, which removes any suspicion that the person running the game picked something they had an answer for.

For game-specific draws, the charades and Pictionary tools carry difficulty ratings this list does not. For a random letter rather than a word, the letter generator draws uniformly and the Scattergories picker draws by playability.

Frequently asked questions

How many words are in the list?

234, curated rather than exhaustive. The tool shows how many survive your filters before you draw.

Why not use a full dictionary?

Because a complete word list is mostly inflections and technical terms nobody would use in a game or a prompt. Curation is what makes the draws usable.

Can a batch contain the same word twice?

No. Batches are drawn without replacement, so ten requested words are ten distinct words.

Are common words more likely than rare ones?

No, every word in the surviving pool has an equal chance. For a frequency-weighted draw, the Scattergories letter picker demonstrates that approach.

Why did my filter leave almost nothing?

Because filters compound. A length band plus a starting substring of 'qu' cuts the 234-word list to five, which the count makes visible.

Can I filter by ending letter or by containing a letter?

No, only length and starting substring. Those are the two filters word games actually use.

What if I ask for more words than the pool holds?

You get the whole pool. A request for twenty from a pool of four returns four.

Can I add my own words?

No. The corpus is curated toward concrete nouns because that is what word games need, and an open list would lose that property immediately.

Is the starting filter case-sensitive?

No. It is matched case-insensitively, so 'Q' and 'q' behave identically.

← Back to all tools