Random Number Generator

About this random number generator

A random number generator (RNG) returns one or more integers drawn uniformly from a closed interval [min, max]. The generator on this page draws from the browser's cryptographic entropy pool (Web Crypto API), so each draw is independent of the previous one and unpredictable to anyone who did not observe the operating system seed. Practical uses include picking a giveaway winner, choosing a row in a spreadsheet for spot-checking, randomizing test data, deciding turn order in a board game, seeding a Monte Carlo simulation, sampling participants for a research study, or just settling an argument without flipping a coin.

Two settings control what you get back: range (min and max) and count. A third toggle, "Allow duplicates," decides whether the draw is with replacement (a duplicate can appear) or without (every number in the output list is unique). Without-replacement draws are equivalent to shuffling a deck and dealing the top N cards, which is what you want for raffle winners or selecting employees for an audit.

How the generator works

Under the hood the page calls crypto.getRandomValues(new Uint32Array(1)) to fetch a 32-bit unsigned integer between 0 and 4,294,967,295, then maps it into the target range using min + (raw % range). For ranges smaller than 2^31 this introduces a tiny modulo bias on the largest residues, undetectable in typical uses; for cryptographic randomness in giant ranges, a rejection-sampling loop would be the textbook fix. The draw repeats until the requested count is reached, with a seen-value map skipping repeats when duplicates are disallowed.

range  = max - min + 1
draw   = min + (crypto32() mod range)
unique = draw and seen[draw] == 0
output = [draw1, draw2, ..., drawN]

• crypto32() = one 32-bit value from crypto.getRandomValues, seeded by the OS entropy pool.
• range = the inclusive count of integers between min and max.
• seen = a hash set that prevents repeats when Allow duplicates is off.

Worked example: pick 5 raffle winners

You run a giveaway with 240 valid entries. Set min = 1, max = 240, count = 5, Allow duplicates off. Click Regenerate. The generator returns five unique integers (for example 73, 12, 198, 207, 41). Match each integer to the row number in your entry spreadsheet to find the winners.

1. Range size: 240 minus 1 plus 1 = 240 unique integers in the pool.
2. Without-replacement draw: the first integer has 240 possibilities, the second has 239, and so on.
3. Probability any specific entry wins: 5 divided by 240 = 2.08 percent.
4. Audit trail: screenshot the output and post it publicly so participants can verify their row number.

Common use cases and recommended settings

Pitfalls to avoid

• Using Math.random for security work. Math.random is a non-cryptographic PRNG (xorshift128+ in V8); attackers can predict its sequence. This tool uses Web Crypto, but if you copy the code, keep the crypto path.
• Confusing without-replacement count caps. If you want 10 unique numbers from 1 to 5, no algorithm can deliver: the range only has 5 integers. The tool surfaces "count exceeds unique range" instead of looping forever.
• Expecting evenly spread results. True randomness is clumpy. Drawing 10 numbers from 1 to 100 often shows two close neighbors and a gap; this is correct uniform behavior, not a bug.
• Re-rolling until you like the result. Each click is fair, but cherry-picking the third roll because you preferred it makes the draw biased and indefensible to anyone who challenges it.
• Forgetting to record the seed for audits. Web Crypto does not expose a seed, so you cannot reproduce a draw. For audit-grade raffles, use a server-side RNG that publishes a hash of the seed in advance.
• Treating small samples as proof of bias. If you draw 1 to 6 ten times and never get a 4, that is normal variance. Run 600 draws and the distribution will flatten.

Related tools on 3Tej

Frequently asked questions

Last updated 2026-05-28.