Pick from a list where some entries are more likely than others. Give each entry a weight β "Alice | 3" is three times as likely as "Ben | 1" β and the tool selects using cumulative weights, so the odds are exactly what you set.
Settings
Recent results
Generated locally in your browser β your settings and results never leave this page.
When an even chance is the wrong answer
Plenty of draws should not be even. A raffle where people bought different numbers of tickets is the obvious case: someone with ten tickets ought to have ten times the chance of someone with one. A loot table with common and rare outcomes needs the rare one to actually be rare. A load test that models real traffic should hit the popular endpoints more often than the obscure ones.
Making these even would be the unfair option. Weighting is what lets the draw reflect the situation it represents.
Weights are ratios
The most common misunderstanding is treating weights as percentages. They are shares. Weights of 3, 1 and 1 give probabilities of 60%, 20% and 20% β the total is five shares, and each entry holds its portion of them.
This is more convenient than percentages in practice, because adding a new entry does not require rewriting everything else. Add “Daniel | 2” to that list and the totals adjust automatically. With percentages, every existing value would need editing to keep the sum at 100.
It also means the same relative odds can be written many ways. Weights of 3 and 1 behave identically to 30 and 10, or 1.5 and 0.5. Use whichever numbers make the intent clearest to a reader.
How the draw is made
The method is cumulative weights. Imagine laying the entries end to end on a line, each occupying a length equal to its weight. Drop a point at random anywhere along the line and see which entry it lands on. An entry three times as long catches the point three times as often.
Implemented directly, that is a running total and a single uniform draw against it. The result is exact rather than approximate β the probabilities really are the ratios the weights describe, with no rounding step where accuracy could leak away.
Drawing several winners
With repeats off, each winner is removed before the next draw and the remaining weights renormalise. This is the right behaviour for a raffle with several prizes: once someone has won, the odds for everybody still in improve.
It also means the second draw is not independent of the first, so the overall probability of a particular person winning something is not simply their share multiplied by the number of prizes. For small lists this can be noticeably different from what people expect, which is worth knowing if you publish odds.
Zero weights and temporary exclusions
Setting a weight to zero removes an entry from the draw without removing it from the list. That is genuinely useful when the same list is reused β a rota where somebody is away this week, a prize draw where an employee is ineligible. Deleting the line risks forgetting to restore it; a zero is visible and reversible.
Disclose weighting when it affects other people
If a draw involving other people is weighted, say so before it runs. A weighted prize draw presented as an even one is misleading even when the weights are entirely reasonable, and the discovery afterwards damages trust more than the weighting itself ever would have.
Where the weights come from something verifiable β tickets purchased, entries submitted, hours contributed β publishing the weight list alongside the result lets people check the arithmetic themselves.
Privacy
Your list is processed in your browser. Nothing is uploaded, stored or placed in the page link.
Checking that weights do what you meant
Weights are easy to write and easy to get subtly wrong, particularly on a long list where the totals are not obvious. An entry with weight 50 among forty entries weighted 1 holds more than half the probability, which may or may not be what you intended when you typed it.
The quickest sanity check is to add up the weights and divide. An entry’s share is its weight over the total, and doing that arithmetic once for the largest entry usually reveals whether the list is doing what you think.
Drawing repeatedly is the other check. If one name comes up far more than it should, or an entry you expected to see never appears, the weights are not what you meant even if each line looks right on its own.
How to use the Weighted Random Picker
- Paste your entries, one per line, adding a weight after a vertical bar β "Alice | 3".
- Turn on weighting if it is not already on.
- Set how many entries to draw.
- Pick, then copy the result or export it.
Frequently asked questions
How do weights work?
A weight is a relative share, not a percentage. "Alice | 3" and "Ben | 1" means Alice is three times as likely as Ben β she holds three of the four shares, so 75%. Weights do not need to add up to anything in particular; only their ratios matter.
What happens to an entry with no weight?
It is treated as weight 1. That means you only need to write weights on the entries that differ from the default, which keeps a long list readable.
Can a weight be zero?
Yes, and it means the entry can never be drawn. That is useful for temporarily excluding somebody without deleting them from your list β set the weight to 0 rather than cutting the line and having to remember it later.
How is the selection actually made?
By cumulative weights. The weights are added into a running total, a random value is drawn uniformly between zero and that total, and the entry whose band contains the value wins. This gives exactly the probabilities the weights describe, with no approximation.
Does drawing several winners change the odds?
Yes, if repeats are off. Once an entry is drawn it is removed and the remaining weights are renormalised, so the odds for everyone else rise. With repeats on, each draw is independent and the odds stay fixed.
When should I not use weights?
Whenever every entry genuinely deserves the same chance. Weighting a draw that people expect to be even is unfair unless it is disclosed, and if a prize draw is weighted, that should be stated in the rules before entries open rather than discovered afterwards.
Can I use decimal weights?
Yes. Weights of 2.5 and 1 behave exactly like 5 and 2. Whole numbers are usually easier to reason about, but there is no requirement for them.
