Random number generator

How to draw your numbers

1. Enter the minimum and maximum. Both ends are included, and negative numbers are fine, so -50 to 50 is a legal range.
2. Set how many numbers you want, up to 10,000 in one draw.
3. Choose whether repeats are allowed. "No repeats" behaves like drawing numbered balls from a drum: each value can appear at most once.
4. Click Generate, then Copy if you need the results elsewhere.

A worked example: a raffle with 250 tickets

You sold tickets numbered 1 through 250 and have five prizes. Set minimum 1, maximum 250, count 5, repeats to No repeats, and click Generate. You get five distinct winning ticket numbers, for example 187 42 250 9 113. Because repeats are forbidden, no ticket can win twice, and because both endpoints are inclusive, ticket 250 has exactly the same chance as ticket 1. If you accidentally ask for more unique numbers than the range holds, say 300 winners from 250 tickets, the tool refuses with a message telling you the range only holds 250 distinct values rather than looping forever or silently allowing duplicates.

Why the draw is genuinely uniform

Every number comes from crypto.getRandomValues, the browser's cryptographic random source, not Math.random. On top of that the code applies rejection sampling: raw 32-bit random values that would introduce modulo bias are discarded and redrawn, so every value in your range is equally likely to the limit of what software can promise. The no-repeats mode is smarter than it looks. For dense draws, like picking 200 of 250 values, it builds the range and runs a partial Fisher-Yates shuffle; for sparse draws, like 5 winners from a million, it draws and rejects duplicates. You never see the difference, but the dense path is what keeps a big draw from grinding as the remaining pool shrinks. Numbers are separated by spaces for short results and commas once you pass 20, purely for readability.

Honest limitations

The generator works in whole numbers only; there is no decimal mode, so for a random price between 1.00 and 5.00 you would generate between 100 and 500 and divide by 100 yourself. Draws are capped at 10,000 numbers, and the dense shuffle path only engages for ranges up to a million values, beyond which unique draws fall back to draw-and-reject, which is fine for small counts and slow if you ask for a huge unique sample from a huge range. There is no draw history and no seed option: you cannot reproduce yesterday's draw, which is a feature for fairness and an annoyance for debugging. Screenshot results you need to prove later.