How the chance of a total is worked out
A fair die with s sides shows each face with probability 1 ÷ s. When n dice are rolled together, every combination of faces is equally likely, and there are sn of them. The chance of a particular total is the share of those combinations that add up to it.
P(total) = ways to make the total ÷ sn
Ways for n dice to make t = sum, over each face f from 1 to s, of the ways for n − 1 dice to make t − f
The second line is how this page counts the ways: it starts with one die, where every total from 1 to s can be made in exactly one way, then adds a die at a time. The counts are kept as exact whole numbers, however large, so the probabilities shown are exact fractions rounded only for display. Two summary figures describe the whole spread:
Average total = n × (s + 1) ÷ 2 + modifier
Standard deviation = √( n × (s² − 1) ÷ 12 )
A modifier is added to the total after the dice are rolled. It slides every result up or down by the same amount and changes nothing about the odds: the chance of 2d6 + 3 coming to 10 is the chance of 2d6 coming to 7.
Worked example: a total of 10 on three six-sided dice
- Count all outcomes. Three dice with six sides give 6 × 6 × 6 = 216 equally likely combinations.
- Start from two dice. Two dice can make 4 in 3 ways, 5 in 4 ways, 6 in 5 ways, 7 in 6 ways, 8 in 5 ways and 9 in 4 ways.
- Add the third die. The third die shows 1 to 6, so the first two must make 9 down to 4. The ways are 4 + 5 + 6 + 5 + 4 + 3 = 27.
- Divide. 27 ÷ 216 = 0.125, a 12.5% chance, or 1 roll in 8.
The average total for three six-sided dice is 3 × 7 ÷ 2 = 10.5, which no roll can show, so 10 and 11 share the top spot with 27 ways each. The standard deviation is √(3 × 35 ÷ 12) = √8.75, about 2.96. At the extremes, 3 and 18 can each be made in only one way, a chance of 0.46%.
One die is flat, several dice are not
A single die gives every face the same chance: 5% for each number on a d20, 8.33% on a d12. Add dice together and the middle totals become far more likely than the ends, because there are more ways to make them. With two six-sided dice there are 36 combinations. Six of them make 7 (16.67%), but only one makes 2 and only one makes 12 (2.78% each).
This is why 2d6 and 1d12 are not interchangeable, even though both top out at 12. On a d12 a 12 comes up three times as often as on 2d6, and a 1 is possible. The more dice in the pool, the more tightly results cluster around the average, and the closer the spread comes to the bell-shaped curve of the normal distribution.
The “this total or higher” figure answers the usual game question. If you need 15 or more on a d20 with a +3 modifier, the die itself must show 12 or more. Nine of its twenty faces do, so the chance is 45%.
How the dice are rolled
Each die is drawn from the cryptographic random source built into your browser, the same source used to make encryption keys. It supplies 32-bit numbers, and a die face is taken from the remainder after dividing by the number of sides. To keep every face exactly equally likely, draws from the uneven top of the range are thrown away and replaced, a method called rejection sampling. For a d20, 4,294,967,280 of the 4,294,967,296 possible draws are used and the last 16 are discarded. The random number generator page explains the method in detail.
- Rolls are independent. The dice have no memory. Three low rolls in a row do not make a high one “due”, and a run of identical results is normal in any long session.
- Nothing is saved or seeded. A roll cannot be replayed, and the address bar does not record your dice, so every visit and every press of the button is a fresh roll.
- d100 and d10. This page numbers faces from 1 to the number of sides. Physical ten-sided dice are usually marked 0 to 9, with 0 read as 10, and a percentile pair showing 00 and 0 is read as 100.
- Limits. Up to 100 dice with up to 1,000 sides each can be rolled. The table of chances is shown when there are 40 or fewer possible totals.