Dice Probability Explained: Two Dice, Sums, and Common Odds
Read in other languages: 日本語
When you roll two six-sided dice, there are 36 equally likely outcomes. The most likely sum is 7, which comes up 6 times out of 36 (1/6, or about 16.67%). Each die shows a single number with probability 1/6. The chance of rolling at least one six in n rolls is 1 − (5/6)^n.
Most dice probability questions come down to one move: count the outcomes you want and divide by the total number of possible outcomes. That only works when every outcome is equally likely, and for fair dice it is. The hard part is counting correctly, and that is where most mistakes happen.
This guide builds the full 36-outcome table and uses it to work out the chance of every sum, the chance of doubles, and the chance of getting at least one six over several rolls. It also covers the average value of a die and how the same ideas carry over to d4, d8, d10, d12, and d20 dice.
The 36 outcomes when you roll two dice
Treat the two dice as different, say a red die and a blue die. The red die has 6 possible faces, and for each of them the blue die has 6. That gives 6 × 6 = 36 ordered outcomes, and each one has probability 1/36.
The table below lists the sum for every pair. Rows are the red die and columns are the blue die.
| Red \ Blue | 1 | 2 | 3 | 4 | 5 | 6 |
|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| 2 | 3 | 4 | 5 | 6 | 7 | 8 |
| 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| 4 | 5 | 6 | 7 | 8 | 9 | 10 |
| 5 | 6 | 7 | 8 | 9 | 10 | 11 |
| 6 | 7 | 8 | 9 | 10 | 11 | 12 |
The key point is that (1, 2) and (2, 1) are two separate outcomes. If you count "a 1 and a 2" as one outcome, you end up with 21 combinations that are not equally likely, and every probability you calculate from them will be wrong. Even when the dice look identical, they are still two separate objects, so (1, 2) and (2, 1) remain distinct outcomes.
Two dice probability: the distribution of sums
Count how many cells in the grid hold each sum and you get the distribution below. Percentages are rounded to two decimal places.
| Sum | Ways | Probability | Percent |
|---|---|---|---|
| 2 | 1 | 1/36 | 2.78% |
| 3 | 2 | 2/36 = 1/18 | 5.56% |
| 4 | 3 | 3/36 = 1/12 | 8.33% |
| 5 | 4 | 4/36 = 1/9 | 11.11% |
| 6 | 5 | 5/36 | 13.89% |
| 7 | 6 | 6/36 = 1/6 | 16.67% |
| 8 | 5 | 5/36 | 13.89% |
| 9 | 4 | 4/36 = 1/9 | 11.11% |
| 10 | 3 | 3/36 = 1/12 | 8.33% |
| 11 | 2 | 2/36 = 1/18 | 5.56% |
| 12 | 1 | 1/36 | 2.78% |
| Total | 36 | 36/36 | 100% |
As a check, the "Ways" column adds up to 1 + 2 + 3 + 4 + 5 + 6 + 5 + 4 + 3 + 2 + 1 = 36.
Why 7 is the most likely sum
A 7 can be made with every possible value of the first die: 1+6, 2+5, 3+4, 4+3, 5+2, and 6+1. Whatever the red die shows, exactly one blue face completes the 7. No other sum has that property. A 2 needs both dice to show 1, and a 12 needs both to show 6, so each has only one way.
That's why the distribution looks like a triangle, with a peak at 7 and equal slopes on both sides. Sums the same distance from 7 have the same probability: 6 and 8, 5 and 9, and so on.
Combining sums: "or," "at least," and "at most"
For a range of sums, add up the ways and divide by 36. This works because a single roll can only have one sum, so the outcomes never overlap.
- 7 or 11: 6 + 2 = 8 ways, so 8/36 = 2/9 ≈ 22.22%.
- 8 or higher: 5 + 4 + 3 + 2 + 1 = 15 ways, so 15/36 = 5/12 ≈ 41.67%.
- 6 or lower: also 15 ways, so 41.67%. Add exactly 7 (6 ways) and you get 15 + 15 + 6 = 36.
- 10 or higher: 3 + 2 + 1 = 6 ways, so 6/36 = 1/6 ≈ 16.67%.
Converting fractions to percentages is just dividing and multiplying by 100. If you want a refresher, see how to calculate percentages.
Probability of rolling doubles
Doubles are the diagonal of the grid: (1,1), (2,2), (3,3), (4,4), (5,5), and (6,6). That's 6 outcomes out of 36, so:
P(doubles) = 6/36 = 1/6 ≈ 16.67%
There's a quicker way to see it. The first die can show anything. The second die then has to match it, which happens with probability 1/6.
A few related numbers:
- One specific double, such as double sixes: 1/36 ≈ 2.78%.
- Not doubles: 30/36 = 5/6 ≈ 83.33%.
- Doubles three rolls in a row: (1/6) × (1/6) × (1/6) = 1/216 ≈ 0.46%.
- At least one six on a single roll of two dice: 11/36 ≈ 30.56%. That's the full sixth row plus the full sixth column, which is 6 + 6 = 12 cells, minus the (6,6) cell that was counted twice.
Chance of at least one six in n rolls
The question "What are the chances I get at least one six?" is easiest to answer from the other direction. Work out the chance of getting no sixes, then subtract that from 1.
Each roll misses a six with probability 5/6. Rolls are independent, meaning one roll doesn't affect the next, so the chance of missing on every one of n rolls is (5/6)^n. That gives:
P(at least one six in n rolls) = 1 − (5/6)^n
Worked example: four rolls
- Chance of no six on one roll: 5/6.
- Chance of no six on four rolls: (5/6)^4 = 625/1296 ≈ 0.4823.
- Subtract from 1: 1 − 625/1296 = 671/1296 ≈ 0.5177.
- Convert to a percentage: about 51.77%.
So four rolls is the smallest number that gives you a better-than-even chance of seeing a six.
Values for common numbers of rolls
| Rolls (n) | No six: (5/6)^n | At least one six |
|---|---|---|
| 1 | 5/6 ≈ 0.8333 | 1/6 ≈ 16.67% |
| 2 | 25/36 ≈ 0.6944 | 11/36 ≈ 30.56% |
| 3 | 125/216 ≈ 0.5787 | 91/216 ≈ 42.13% |
| 4 | 625/1296 ≈ 0.4823 | 671/1296 ≈ 51.77% |
| 5 | 3125/7776 ≈ 0.4019 | 4651/7776 ≈ 59.81% |
| 6 | 15625/46656 ≈ 0.3349 | 31031/46656 ≈ 66.51% |
| 10 | ≈ 0.1615 | ≈ 83.85% |
| 13 | ≈ 0.0935 | ≈ 90.65% |
| 17 | ≈ 0.0451 | ≈ 95.49% |
| 20 | ≈ 0.0261 | ≈ 97.39% |
You need 13 rolls to pass 90% and 17 rolls to pass 95%. No number of rolls makes it certain. The chance of a miss keeps shrinking but never reaches zero.
The same formula works for any target. Swap 1/6 for the chance of success on one try (call it p), and the chance of at least one success in n tries is 1 − (1 − p)^n.
Why you can't just add 1/6 each time
It's tempting to say two rolls give 1/6 + 1/6 = 1/3 and six rolls give 6/6, a sure thing. Both are wrong. Adding counts the cases with more than one six twice. For two rolls, the outcome (6, 6) gets counted once for each die, which is why the right answer is 11/36 and not 12/36.
Over six rolls the double counting grows, and the real answer is about 66.51%, not 100%.
The double-sixes puzzle
There's a classic version of this problem tied to the Chevalier de Méré, which Blaise Pascal and Pierre de Fermat discussed in letters in the 1600s. Is it worth betting on at least one double six in 24 rolls of two dice? Use the formula with p = 1/36:
1 − (35/36)^24 ≈ 1 − 0.5086 ≈ 49.14%
De Méré reasoned that 4 rolls with 6 faces should match 24 rolls with 36 outcomes, since both ratios are the same (4/6 = 24/36). But the four-roll bet wins about 51.77% of the time, while the 24-roll bet wins only 49.14%. So the first bet comes out slightly ahead over time and the second slightly behind. To get above 50% on double sixes you need 25 rolls: 1 − (35/36)^25 ≈ 50.55%.
Expected value of a die
The expected value is the long-run average of many rolls. For one fair d6, multiply each face by its probability and add the results:
(1 + 2 + 3 + 4 + 5 + 6) × 1/6 = 21/6 = 3.5
No single roll can show 3.5. It's the average you approach after many rolls, not a likely result. Roll 600 times and you should expect a total close to 600 × 3.5 = 2,100, though not exactly that.
Expected values add up, even when the dice are different. Two dice average 3.5 + 3.5 = 7, which matches the peak of the sum table. Three dice average 10.5. Note that the most likely result and the average don't always line up so neatly. With three d6, the sums 10 and 11 tie for most likely, each at 27/216 = 12.5%.
Expected value also lets you judge a game quickly: multiply each outcome's payout by its probability and add the results. Say you pay $1 to roll two dice and win $5 if the sum is 7. The expected payout is 5 × 1/6 ≈ $0.83, so on average you lose about 17 cents per play.
Probability for d4, d8, d10, d12, and d20 dice
Tabletop games use other polyhedral dice, named "d" plus the number of faces. For a fair die with faces 1 to n, each face has probability 1/n and the expected value is (n + 1)/2.
| Die | Faces | Chance of any one face | Expected value |
|---|---|---|---|
| d4 | 4 | 1/4 = 25% | 2.5 |
| d6 | 6 | 1/6 ≈ 16.67% | 3.5 |
| d8 | 8 | 1/8 = 12.5% | 4.5 |
| d10 | 10 | 1/10 = 10% | 5.5 |
| d12 | 12 | 1/12 ≈ 8.33% | 6.5 |
| d20 | 20 | 1/20 = 5% | 10.5 |
Many d10s are numbered 0–9 rather than 1–10. In that case the average is 4.5, so check how your game reads the 0.
Everything above still applies. Two dice with n faces each have n × n outcomes, and the sums form a triangle that peaks at n + 1. For two d20s that's 400 outcomes and a most likely sum of 21, at 20/400 = 5%. The at-least-once formula works too. Roll a d20 twice and the chance of at least one 20 is 1 − (19/20)^2 = 39/400 = 9.75%.
You can check these patterns yourself with the dice roller, which rolls up to six dice from d4 to d20 and shows totals and stats. Over a few hundred rolls the frequencies drift toward the values in these tables. Over a few dozen they can look quite uneven, and that's normal.
Common mistakes with dice odds
- Counting unordered pairs. "A 3 and a 4" is two outcomes, (3,4) and (4,3), so it's twice as likely as "two 3s."
- Thinking a number is "due." Dice have no memory. After five rolls with no six, the next roll is still 1/6. This is the gambler's fallacy, which is explained in more depth in coin flip probability.
- Mixing up probability and odds. The probability of a 7 is 6/36. The odds compare successes to failures: 6 to 30, or 1 to 5 in favor (5 to 1 against). A probability of 1/6 and odds of 1:5 describe the same chance.
- Thinking digital dice are rigged or "streaky." Most software uses a pseudo-random number generator, an algorithm that produces numbers that look random. A well-built one gives each face an equal 1/6 share, and streaks happen just as they do with physical dice. For more on how that works, see how random number generators work.
Dice probability quick reference
- One die, any face: 1/6 ≈ 16.67%
- Two dice, total outcomes: 36
- Most likely sum (2d6): 7, at 6/36 ≈ 16.67%
- Sum of 2 or 12: 1/36 ≈ 2.78% each
- Doubles: 6/36 = 1/6 ≈ 16.67%
- Double sixes: 1/36 ≈ 2.78%
- At least one six, two dice, one roll: 11/36 ≈ 30.56%
- At least one six in n rolls: 1 − (5/6)^n. Four rolls ≈ 51.77%, six rolls ≈ 66.51%
- General "at least once": 1 − (1 − p)^n
- Average of one d6: 3.5. Average of n-sided die: (n + 1)/2
- Before you calculate: count ordered outcomes, make sure they're equally likely, and for "at least one" questions, calculate the complement first.
Frequently Asked Questions
On average, how many rolls does it take to get a six?
Six rolls. When each try succeeds with probability p, the average number of tries until the first success is 1/p, and 1 ÷ (1/6) = 6. Individual runs vary a lot. About a third of the time you still won't have a six after six rolls.
What is the probability of rolling the same number on three dice?
Three dice have 6 × 6 × 6 = 216 outcomes, and 6 of them are triples: (1,1,1) through (6,6,6). That makes the probability 6/216 = 1/36, or about 2.78%. The chance of one specific triple, such as three sixes, is 1/216, or about 0.46%.
What are the chances of rolling a 12 twice in a row with two dice?
A 12 has probability 1/36 on each roll, and the rolls are independent, so you multiply: 1/36 × 1/36 = 1/1296. That's about 0.077%, or roughly 1 in 1,300 attempts.
Does rolling both dice at once give different odds than rolling one die twice?
No. As long as the dice are fair and each roll is independent, rolling two dice together and rolling one die twice produce the same 36 equally likely ordered outcomes. The timing has no effect on the probabilities.
How can I tell if a physical die might be unfair?
Roll it many times and compare how often each face comes up against the expected 1/6. Small samples swing widely, so a few dozen rolls tell you very little. Hundreds of rolls give a clearer picture. A formal check uses a statistical test such as the chi-square goodness-of-fit test.