Roll dice of any size — d4 to d100 — with an optional modifier, session statistics, and the exact probability distribution for your dice combination. Runs entirely in your browser — your numbers never leave this device.
| Sum | Ways | Probability |
|---|---|---|
| 2 | 1 | 2.78% |
| 3 | 2 | 5.56% |
| 4 | 3 | 8.33% |
| 5 | 4 | 11.11% |
| 6 | 5 | 13.89% |
| 7 | 6 | 16.67% |
| 8 | 5 | 13.89% |
| 9 | 4 | 11.11% |
| 10 | 3 | 8.33% |
| 11 | 2 | 5.56% |
| 12 | 1 | 2.78% |
Dice are drawn from the browser's cryptographic random source with rejection sampling, so every face is equally likely — never Math.random.
| Sum | Ways | Probability |
|---|---|---|
| 2 | 1 | 2.78% |
| 3 | 2 | 5.56% |
| 4 | 3 | 8.33% |
| 5 | 4 | 11.11% |
| 6 | 5 | 13.89% |
| 7 | 6 | 16.67% |
| 8 | 5 | 13.89% |
| 9 | 4 | 11.11% |
| 10 | 3 | 8.33% |
| 11 | 2 | 5.56% |
| 12 | 1 | 2.78% |
Choose how many dice to roll and how many sides each one has — the standard polyhedral set from d4 to d20, plus d100 — add a modifier if your game calls for one, and roll. Each face is drawn from the browser's cryptographic random source with rejection sampling, so every face of every die is equally likely, and the total is the dice sum plus your modifier. The session panel keeps count of your throws for the current dice setup, with the mean, lowest and highest sums, so a run of bad luck is visible as a run — and changing the dice restarts the session, since a 2d6 average and a 1d20 average are not the same statistic.
The substance under the roller is the odds table. The distribution of a dice sum is not flat, and it is not a matter of simulation or opinion: it is counted exactly. Two six-sided dice have 36 equally likely outcomes; the table counts how many of them produce each sum by convolving the single-die counts — rolling the one-die distribution through itself once per die. That is why 7 is the most likely 2d6 result, with 6 of the 36 outcomes (16.67%), while 2 and 12 each have exactly one outcome (2.78%). Three dice sharpen the bell further: 10 and 11 are the peaks of 3d6 at 27 of 216 outcomes each (12.5%), and the extremes 3 and 18 sit at 1 in 216. The table on the tool rebuilds itself for whatever combination you select, whenever the number of possible sums is small enough to print.
Read the table before a game, not just during one. It explains the bets and the character builds: a weapon or a rule that adds a flat modifier shifts the whole distribution without changing its shape, while adding a die changes the shape itself, pulling results toward the middle. And the honest warning that belongs on every dice page: past rolls change nothing. Each throw is independent — a die that has rolled low five times has the same chance of rolling low on the sixth as it ever did, and the session mean drifting toward the theoretical average over many throws is the law of large numbers doing its slow work, not the dice evening any score.
Because totals are made of combinations, and the middle totals have more of them. A sum of 7 on 2d6 is made six ways; a sum of 2 only one way, double one. The dice themselves are uniform — each of the 36 face combinations is equally likely — but sums inherit the count of their combinations, which is what the convolution table on this page counts exactly.
It shifts them without reshaping them. Adding +3 to 2d6 moves every outcome up by three, so the most likely total becomes 10 instead of 7 with the same 16.67% probability, and the range becomes 5–15. Adding another die instead would change the shape — the bell gets taller and narrower relative to its range — which is why games treat 'an extra die' and 'a plus' as very different rewards.
This one draws each face from crypto.getRandomValues with rejection sampling, the same unbiased method the Random Number Generator on this site documents — every face is equally likely and past faces cannot influence future ones. No browser tool can be witnessed the way physical dice at a table can, so for a high-stakes game the table's own dice remain the gold standard.
Half of (sides + 1). A d6 averages 3.5, a d20 averages 10.5, a d100 averages 50.5. For several dice, add the averages — 3d6 averages 10.5, the centre of its 3–18 range, which sits between the twin peaks at 10 and 11 shown in the worked example.
Because a mean only means something for one distribution. Mixing 2d6 throws (average 7) with 1d20 throws (average 10.5) in one running average would produce a number that describes neither. The session panel therefore tracks throws for the current dice setup only; the Roll button adds to it, and changing the count or sides starts a fresh session.
More in Everyday Tools & Generators · Browse all tools