Skip to content
uencoder
TUTORIAL

How to Generate Random Numbers

Generate cryptographically secure random numbers, integers, and bytes.

Problem

You need random numbers for security purposes, testing, or simulation, and must ensure proper randomness.

Prerequisites

  • Understanding of PRNG vs CSPRNG
  • Knowledge of random number ranges

Steps

  1. 1. Choose the right generator

    For security purposes (tokens, keys, passwords), use a cryptographically secure PRNG (CSPRNG): crypto.getRandomValues in browsers, crypto.randomBytes in Node.js. For non-security uses (testing, simulation), Math.random is faster but not secure. Never use Math.random for security.

    Input
    purpose: security token
    Output
    use CSPRNG
  2. 2. Generate within the desired range

    For a random integer in [min, max], generate random bytes and map to the range. Avoid modulo bias: use rejection sampling or generate enough bits to cover the range uniformly. For random floats in [0, 1), divide a large random integer by its maximum value.

    Input
    range: [1, 100]
    Output
    42 (uniform)
  3. 3. Generate other random types

    For random bytes (e.g., for a key), generate the required number of bytes directly. For random hex strings, convert the bytes to hex. For UUIDs, use the dedicated UUID generator. For random selections from an array, use a random index.

    Input
    32 random bytes
    Output
    a1b2c3... (64 hex chars)

Expected result

Random values generated with the appropriate level of security and uniformity.

Common errors

Modulo bias when using % to range
Using randomNumber % range produces biased results when the range doesn't evenly divide the generator's range. Use rejection sampling or crypto.randomInt for unbiased results.
Using Math.random for security
Math.random is predictable and not cryptographically secure. Always use crypto.getRandomValues or crypto.randomBytes for security-sensitive randomness.

Troubleshooting

  • If random values seem predictable, verify you are using a CSPRNG.
  • For large numbers of random values, check the distribution for uniformity.

Security limits

  • All random generation is local. No random values are transmitted.
  • CSPRNG output is unpredictable and suitable for security tokens, keys, and nonces.

Related tools

Related tasks

Ready to run it?

References