A random number generator answers a narrow question: give me a number in this range, with every number equally likely. That sounds trivial until you try to do it fairly. People asked for a "random" number between 1 and 10 pick 7 far more often than chance allows, and naive code has its own bias. This generator draws from the operating system’s cryptographic random source through your browser and removes the rounding bias that a plain remainder introduces, so every value in the range really is equally likely.
How the numbers are drawn
- Each draw asks the Web Crypto API (
crypto.getRandomValues) for a 32-bit random integer. This is the same source password managers and encryption keys use, not a seeded pseudo-random sequence. - Values that fall in the last, incomplete block of the 32-bit space are thrown away and re-drawn. Without this step, taking the remainder makes the lowest numbers in the range slightly more likely: the classic modulo bias.
- The result is mapped onto your range: from + (random value modulo the size of the range).
- For unique draws, a number already picked is discarded and another is drawn, so no value can repeat.
Example
A range of 1 to 49 holds 49 numbers. Draw 6 of them without repeats and there are 13,983,816 possible tickets, which is where the familiar 1-in-14-million lottery odds come from. Any one number you name has a 6 in 49 chance, about 12.2%, of appearing somewhere in your six.
Repeats change the odds
With duplicates allowed, each draw is independent, so the chance that a given number shows up at least once in n draws is 1 − (1 − 1/range)ⁿ. Rolling a six-sided die twice gives any particular face a 30.6% chance of appearing, not 33%. With unique draws the arithmetic is simpler: n numbers out of a range of N means every number has an n/N chance.
What this is not
For a random string of characters rather than numbers, the password generator draws from the same source. This is a fair draw, not a prediction or a strategy. Past draws tell you nothing about the next one, and no arrangement of numbers improves lottery odds. If you need a draw that other people must be able to audit, for a prize promotion or a study, record the seed method and the witnesses rather than relying on any website: nothing here is logged, so there is no record to show afterwards.
Frequently asked questions
Are these numbers really random?
They come from your device’s cryptographic random number generator, the same source used for encryption keys, and the mapping onto your range is done with rejection sampling so no value is favoured. That is as close to fair as software gets.
How do I roll dice or flip a coin?
A six-sided die is 1 to 6, one number. Two dice are 1 to 6, two numbers, duplicates allowed. A coin flip is 1 to 2. The table above lists the settings for the usual cases.
How do I pick lottery numbers?
Set the range to the game’s numbers, for example 1 to 49, ask for six, and choose unique only so nothing repeats. Any set of numbers is exactly as likely as any other.
Can I get the same number twice?
Only if duplicates are allowed. Choose unique only for raffles, team picks and lottery lines, where a repeat would be wrong.
Is anything sent to a server?
No. The draw happens in your browser and the numbers are never transmitted or stored, so we cannot reproduce or verify a past draw for you.
Sources
- MDN: Crypto.getRandomValues() — the cryptographically strong source used for every draw
- NIST SP 800-90A Rev. 1: Recommendation for Random Number Generation Using Deterministic Random Bit Generators
By Calcelate Team. Formula from the sources above.
- 2026-09-15 · Formula and texts checked