Random Number Generator
Pick a range, pick a count, get numbers that are genuinely evenly spread. Every draw on this page is powered by your browser's cryptographic random source and corrected with rejection sampling, so no value in your range is quietly more likely than any other.
Need a password instead of a number? Generate one locally with PassGen — it uses the same cryptographic source and never sends anything anywhere.
How these numbers are drawn
Most quick generators do something like Math.floor(Math.random() * 100). That has two weaknesses. First, Math.random() is a fast pseudo-random generator that was never designed to be unpredictable. Second, mapping a random value onto a range with a remainder (modulo) operation biases the result whenever the draw space is not an exact multiple of the range: some outcomes get one extra way to occur.
RandomNum avoids both problems:
Entropy source. Bytes come from crypto.getRandomValues, the browser's cryptographically secure random number generator (CSPRNG), which is seeded from your operating system's entropy pool.
Rejection sampling. To draw from 1–10, the engine takes one random byte (256 possible values), computes the largest multiple of 10 that fits — 250 — and simply redraws any byte of 250 or above. The remaining 250 values divide into exactly 25 complete sets of 10, so every number from 1 to 10 has precisely 25 ways to occur. No favourite, no bias. The same method scales to any range: enough whole bytes are drawn to cover the range, and only the incomplete final block is ever rejected.
Verified, not just claimed. This site's engine was distribution-tested before launch: 100,000 draws across 1–10 landed within 1.67% of the expected 10,000 per value (actual counts ran from 9,853 to 10,167), and 60,000 rolls of a 20-sided die stayed within 3.90% of even per face (face counts from 2,883 to 3,087 against 3,000 expected) — ordinary statistical noise for samples that size, with no value systematically favoured.
Honest limits
A CSPRNG is still a pseudo-random generator: extremely well-seeded and unpredictable in practice, but algorithmic rather than truly random in the physical sense. That is the right tool for games, raffles among friends, sampling, testing and everyday decisions. It is not a certified drawing machine — official lotteries use audited hardware, and cryptographic keys should come from your platform's key management, not a web page. We would rather say that plainly than oversell the word "random".
Popular ranges
Random number 1–10 · Random number 1–100 · Random number 1–1000 · Dice roller · Coin flip · Name picker
FAQ
- How many numbers can I generate at once?
- Up to 10,000 per batch. That cap keeps your browser responsive; just run another batch if you need more.
- What does "No duplicates" do?
- It keeps drawing until it has the requested count of distinct values — the right setting for raffles and lottery-style picks. If you ask for more unique values than the range contains, the tool will tell you instead of looping forever.
- Is Math.random() really biased?
- The bias from a plain modulo mapping is tiny for small ranges, but it is real and grows with the range size. Rejection sampling removes it entirely, at the cost of occasionally discarding a draw — a fair trade.
- Are my numbers sent to a server?
- No. Generation happens entirely in this tab, using your device's cryptographic random source. Nothing you generate is uploaded, logged or stored.
- Can I use this for an official lottery or giveaway?
- For informal draws, yes. For anything regulated or high-stakes, use a certified drawing process — a browser CSPRNG is unpredictable, but it is not audited hardware.