Dice Probability Calculator

Compute exact dice probabilities for any number of dice and sides.

Science Exact probabilities Distribution chart
Rate this calculator · 5.0 (1)

Dice Probability

d4-d20 supported, exact enumeration, distribution chart

Instructions — Dice Probability Calculator

1

Pick the die type

Choose the die you are rolling: d4, d6, d8, d10, d12, or d20. A standard cubic die is d6. Tabletop role-playing games (D&D, Pathfinder) use the full set. The calculator assumes a fair die where every face has probability 1/s.

2

Set the number of dice and target

Enter how many dice you roll (1-10) and the target sum. With two d6, the sums run from 2 to 12. With three d6, from 3 to 18. The calculator clamps targets outside the achievable range to a probability of zero.

3

Pick a comparison

"Exact sum" returns P(S = target). "At least" returns the upper tail P(S ≥ target), useful for ability checks where you need to roll above a difficulty class. "At most" returns the lower tail, useful for failure thresholds and damage rolls.

Output explained: The calculator reports the favorable outcomes over total outcomes (e.g. 6/36 for 2d6 summing to 7), the probability as a percentage, and the full distribution chart so you can see where your target falls on the curve.
Method: Full polynomial-convolution enumeration. Every possible outcome is counted; this is not Monte Carlo simulation. For 10d20 the answer is exact to the last digit, computed in well under a millisecond.

Formulas

Dice probability rests on counting equally likely outcomes. For one fair die with s sides, each face has probability 1/s. For n dice, total outcomes is s^n, and the count of ways to hit any particular sum is the n-fold convolution of the single-die distribution.

Single die
$$ P(X = k) = \frac{1}{s},\ \ k \in \{1, \dots, s\} $$
Each face of a fair s-sided die is equally likely. For a d6, P(X = 3) = 1/6 = 16.67%. For a d20, each face = 5%.
Total outcomes for n dice
$$ N = s^n $$
For 2d6, N = 36. For 3d6, N = 216. For 5d20, N = 3,200,000. The calculator enumerates this space exactly.
P(exact sum)
$$ P(S = t) = \frac{c_t}{s^n} $$
c_t is the number of ordered tuples (x_1,..., x_n) with x_i in {1..s} summing to t. For 2d6 hitting 7, c_t = 6 and P = 6/36 = 16.67%.
Expected value
$$ E[S] = n \cdot \frac{s + 1}{2} $$
Mean of one die is (s+1)/2. For 2d6, E[S] = 7. For 3d20, E[S] = 31.5. Linearity of expectation does not require independence.
Variance
$$ \mathrm{Var}(S) = n \cdot \frac{s^2 - 1}{12} $$
Per-die variance is (s^2-1)/12. For 2d6, Var = 5.833 and SD ≈ 2.415. Larger dice have higher per-roll variance.
Cumulative probability
$$ P(S \ge t) = \sum_{k=t}^{n s} P(S = k) $$
At-least tail. Useful for D&D ability checks where rolling at or above a target succeeds. At-most tail is the symmetric lower sum.

Reference

2d6 sum distribution
SumWaysP(sum)
212.78%
325.56%
438.33%
5411.11%
6513.89%
7616.67%
8513.89%
9411.11%
1038.33%
1125.56%
1212.78%

Single-die statistics

Mean E[X]
DieMean
d42.5
d63.5
d84.5
d105.5
d126.5
d2010.5
Standard deviation
DieSD
d41.118
d61.708
d82.291
d102.872
d123.452
d205.766

Article — Dice Probability Calculator

Dice Probability Calculator

Dice probability is the chance of a specific outcome when rolling one or more fair dice. For two six-sided dice, the chance of rolling a seven is 6/36 = 16.67%, the most likely sum. The exact formula for n s-sided dice is P(S = t) = c_t / s^n, where c_t counts the ordered tuples summing to t.

This page covers the math behind every dice roll: how to count outcomes for any number of dice, why the sum distribution bends into a bell curve, what to do when you need an at-least or at-most probability, and where common intuition fails. Every value is computed by exact enumeration in the calculator, not a Monte Carlo approximation.

What is dice probability?

Dice probability is the ratio of favorable outcomes to total outcomes when rolling fair dice. A fair die has equal probability on each face. For a standard six-sided die, every face has chance 1/6 ≈ 16.67%. For a twenty-sided die used in Dungeons & Dragons, each face has chance 1/20 = 5%.

Two ideas anchor the rest: the sample space (every outcome that can happen) and the event (the subset you care about). For two d6, the sample space has 36 ordered pairs. If you care about pairs summing to 7, the event has 6 elements. The probability is 6/36.

Did you know

Galileo wrote one of the first known dice probability essays around 1620, settling a gambling dispute. He showed that rolling a 10 with three dice is slightly more likely than rolling a 9, even though both can be made in six unordered ways. The ordered count, which is what matters, is 27 ways for 10 versus 25 ways for 9 out of 216 total.

Dice probability for one die

For one fair s-sided die, P(X = k) = 1/s for any k in {1,..., s}. That uniform shape is the building block for every multi-die calculation. The expected value is (s + 1) / 2: 2.5 for a d4, 3.5 for a d6, 10.5 for a d20.

Variance for a single die equals (s^2 − 1) / 12. For a d6 that is 35/12 ≈ 2.917; the standard deviation is the square root, about 1.708. A d20 has variance 33.25 and SD about 5.77, which is why a single d20 roll feels swingier than a single d6.

Dice probability for the sum of multiple dice

For n dice the total number of outcomes is s^n. For 2d6, that is 36. For 3d6, 216. For 4d6, 1,296. To get the probability of a sum, count the ordered tuples that hit the target and divide by s^n. With 2d6 the count by sum is the familiar triangle: 1, 2, 3, 4, 5, 6, 5, 4, 3, 2, 1 for sums 2 through 12.

For more than two dice, the count is harder to do by hand. The calculator builds the full distribution by convolution: multiply the single-die generating polynomial (x + x^2 +... + x^s) by itself n times and read off the coefficients. That gives the exact count for every sum in a few microseconds, even for 10d20.

Counts for 2d6 by sum
2 → 1 3 → 2
4 → 3 5 → 4
6 → 5 7 → 6
8 → 5 9 → 4
10 → 3 11 → 2
12 → 1 total 36

Expected value and variance of dice rolls

Linearity of expectation makes the mean of a sum easy: E[S] = n * E[X] = n * (s + 1) / 2. Two d6 average to 7. Three d20 average to 31.5. For 4d6 keep highest three (a common D&D ability score roll) the mean is about 12.24, which is why the standard scores tend to cluster between 11 and 14.

Variance also adds for independent dice: Var(S) = n * (s^2 − 1) / 12. The standard deviation is the square root and grows like √n, not like n. That is the source of the bell shape: relative to the mean, multi-die sums concentrate more tightly as n increases.

Dice probability in tabletop RPGs

In D&D 5th edition, an ability check rolls 1d20 + modifier against a difficulty class. The chance of meeting or exceeding DC X with modifier M is P(d20 ≥ X − M). For modifier +5 against DC 15, that is P(d20 ≥ 10) = 11/20 = 55%. Advantage rolls 2d20 and keeps the higher; the success probability becomes 1 − ((X − M − 1)/20)^2 ≈ 79.75% for the same example.

For damage, the calculator handles the at-least and at-most modes you need. A fireball does 8d6 damage; the expected value is 28 and the SD is about 4.83. The chance of rolling 35 or higher on 8d6 is about 7.65% — a useful number for game balance and encounter design.

Tip

For the chance of "at least one success" across n independent trials, the cleanest formula is the complement: 1 − (1 − p)^n. Rolling a 20 on a single d20 has p = 0.05; the chance of at least one 20 across 10 attempts is 1 − 0.95^10 ≈ 40.13%.

Common dice probability mistakes

The most common error is the gambler's fallacy: assuming that because a 6 has not come up recently, it is "due." Dice have no memory. Each roll is independent. After ten non-sixes, the next roll has the same 1/6 chance as the first.

Another trap is counting unordered combinations instead of ordered ones. For 2d6 summing to 3, there are two ordered pairs (1,2) and (2,1) — not one. The sample space treats the dice as distinguishable, so always count ordered tuples. Galileo's three-dice paradox is exactly this trap.

Independence breaks for non-replacement

The dice probability formulas on this page assume each die is independent. If you are drawing dice from a finite set without replacement (rare, but it happens in some board games), the calculations change. For ordinary dice rolling, independence always holds.

Monte Carlo simulation vs. exact enumeration

Many online dice calculators use Monte Carlo: roll the dice 100,000 times in code and count successes. That works, but it gives an approximation with sampling error roughly 1/√N. For 100k trials, error is about 0.3%. The calculator on this page uses exact enumeration via polynomial convolution, so the answer is the exact rational probability with no sampling noise.

A dice probability cheat sheet

  • 1d6 specific face = 1/6 ≈ 16.67%
  • 1d20 natural 20 = 1/20 = 5%
  • 2d6 sum of 7 = 6/36 ≈ 16.67% (most likely sum)
  • 2d6 doubles = 6/36 ≈ 16.67%
  • 3d6 sum of 10 or 11 = 27/216 each ≈ 12.50%
  • 4d6 drop lowest mean ≈ 12.24
  • At least one 20 in n rolls = 1 − (19/20)^n
  • Variance of nd6 = 35n/12, SD = √(35n/12)

FAQ

P(sum = 7) = 6/36 = 16.67%. Six ordered pairs add to seven: (1,6), (2,5), (3,4), (4,3), (5,2), (6,1). Out of 36 total equally likely outcomes for two d6, that gives 1/6. Seven is the most likely sum on 2d6 and the center of the triangular distribution.
Count favorable outcomes divided by s^n total outcomes. For three d6 summing to exactly 10, enumerate every ordered triple in {1..6}^3 that sums to 10. There are 27 such triples out of 216, giving P = 27/216 = 12.5%. The calculator above does this by polynomial convolution, which is faster than nested loops but mathematically identical.
E[S] = n * (s+1)/2. For 2d6: 2 * 3.5 = 7. For 3d20: 3 * 10.5 = 31.5. For 4d6 dropping the lowest (D&D ability scores): about 12.24. The expected value of a single fair die is the average of 1 through s, which is (s+1)/2.
The Central Limit Theorem says sums of independent random variables tend toward a normal distribution. One die is uniform (flat). Two dice form a triangle. Three or more dice start to round into a bell shape. By 10 or more dice the distribution is nearly indistinguishable from a Gaussian with mean n(s+1)/2 and variance n(s^2-1)/12.
6 out of 36, or 16.67%. The doubles are (1,1), (2,2), (3,3), (4,4), (5,5), (6,6). On a single die, P(specific face) = 1/6. Two dice landing on the same face independently is 6 * (1/6) * (1/6) = 1/6. This is the same probability as rolling exactly a 7.
Exactly 5%, or 1 in 20. A fair twenty-sided die has uniform probability 1/20 = 5% for each face. The probability of at least one nat 20 across n rolls is 1 - (19/20)^n. For two rolls that is 9.75%; for ten rolls about 40.13%; for twenty rolls about 64.15%.
Exact enumeration, not Monte Carlo. The calculator builds the full distribution by convolving the single-die distribution n times. For 10d20 it counts 10,240,000,000 outcomes implicitly via the polynomial method and gives the exact rational probability, not an approximation.
Var(X) = 35/12 ≈ 2.917, SD ≈ 1.708. For n independent d6, variance is 35n/12 and SD is the square root. For 2d6: Var = 5.833, SD ≈ 2.415. For 10d6: Var ≈ 29.17, SD ≈ 5.40. Higher variance means rolls spread further from the mean.