What people actually use random numbers for

Six categories cover almost every real request, and they need different kinds of randomness.

Random numbers Published October 20, 2026 2 min read

The six categories

Use	Typical range	With or without repeats
Fair draw / raffle	1 to entry count	Without
Sampling for review	1 to record count	Without
Game mechanics	Dice, d20, cards	Depends on the dice
Test data	Any range or length	Either, usually with
Task or seat allocation	1 to person count	Without
Decisions and tie-breaks	1 to option count	N/A

Without replacement is the one people get wrong

Drawing winners one at a time and re-rolling on a duplicate is the manual approach, and it has two problems: it wastes attempts, and it makes the process look arbitrary even when it is fair. Sampling without replacement — a partial shuffle — is both correct and easier to explain.

The distinction also changes the maths. With replacement, each draw has the same probability. Without, the pool shrinks, so probabilities shift after every pick.

Where randomness should not be used

Anything that needs to be reproducible: test fixtures that must match across runs, A/B assignment that must be stable for a user, or a deterministic simulation. In those cases you want a seeded generator, not a random one — the same seed reproduces the same sequence.

And anything security-relevant should use a cryptographic source, not a convenience page. A random number for a raffle and a random number for a token look identical but have opposite requirements.

Frequently asked questions

What is the difference between a draw and a pick?

A draw usually implies order and no repeats; a pick implies a single independent choice. For multiple winners, you want the draw behaviour.

Can I use a random number generator for a giveaway?

For a casual one, yes. For anything with legal weight, document the method and the entry list before drawing, and keep the ordered result.

Should test data be random or seeded?

Seeded, if the tests depend on it. Random values that change every run make failures hard to reproduce.

Tools that go with this guide

Random Number Generator Any range, any count, uniform distribution. Both ends included. Random Picker Paste your list, one item per line, and draw a winner. Random Draw Sequential draws for multiple prizes, without the re-roll problem. Multiple Random Numbers A list of numbers at once, with or without repeats.

More guides

Passwords How long does it take to crack a password? The answer depends almost entirely on two things: how the password is stored, and how long it is. Passwords What actually makes a password strong Four properties matter. Most of the advice people repeat is about the least important one. Passwords Password generator vs password manager: which do you need? They solve different halves of the same problem, and using one without the other barely helps. QR codes Why won't my QR code scan? Almost every failed scan comes down to one of eight causes. Here is how to tell which.