Start with one fair die
A fair six-sided die has six faces that are equally likely, so each face has a probability of 1/6, about 16.7 percent. Every other calculation in this guide is built from that one fact plus a few rules for combining it.
The most important idea is that rolls are independent. The die has no memory, so a long run without a six does not make a six more likely on the next roll, and a run of sixes does not make it less likely. Players often feel otherwise, which is why the idea is worth stating plainly. A related misunderstanding is covered under the hot hand belief, which looks at why streaks feel meaningful even when they are not.
Dice come in other sizes, and the same logic applies. A four-sided die gives each face a 1/4 chance, a ten-sided die 1/10, a twenty-sided die 1/20. A simple target is just a count of favourable faces divided by the total number of faces. Needing 15 or more on a twenty-sided die means six favourable faces (15 through 20), so the chance is 6/20, or 30 percent.
Two dice and the shape of a sum
Roll two six-sided dice and add them, and the outcomes are no longer equally likely. There are 36 ordered pairs, each with probability 1/36, but several pairs share a total. A total of 7 can be made six ways, while 2 and 12 can each be made only one way. This is why totals bunch up around the middle, a pattern that appears again in the normal distribution as more dice are added.
Three dice show the effect more strongly. There are 6 × 6 × 6 = 216 ordered outcomes, using the multiplication principle: multiply the number of options at each step. Totals run from 3 to 18, and 10 and 11 are the most common, with 27 outcomes each, so each has a probability of 27/216, or 1/8. A total of 18 requires three sixes and has a probability of 1/216. The more dice are summed, the more the extremes thin out.
The reason sums behave this way is that you are counting combinations, not totals. The totals are not equal in how many ways they can arise. Whenever a game asks for a sum, the first question to ask is how many of the equally likely outcomes produce it.
“At least one” and the complement shortcut
Many game rules ask whether something happens at least once. Counting every way to get one or more sixes is slow, but the opposite event is easy: no sixes at all. This is complement counting, and it works like this.
- The chance of no six on one die is 5/6.
- For two dice, both must miss, so the chance is (5/6) × (5/6) = 25/36.
- The chance of at least one six on two dice is therefore 1 − 25/36 = 11/36, about 30.6 percent.
For four dice, no six has probability 625/1296, so at least one six is 671/1296, a little over 51.8 percent. Notice that rolling six dice does not guarantee a six; the chance of none is still (5/6)^6, roughly 33.5 percent. Adding dice raises the chance steadily, but never to certainty.
Counting successes: pools and thresholds
Some games ask you to roll a handful of dice and count how many meet a target, such as showing a 5 or 6. Each die is a separate trial with the same success chance, here 1/3. With three dice, the number of successes follows a binomial distribution:
| Successes | Probability |
|---|---|
| 0 | 8/27 |
| 1 | 12/27 |
| 2 | 6/27 |
| 3 | 1/27 |
These add to 27/27, which is a useful check. The expected number of successes is three dice times 1/3, which is exactly one. Making the threshold easier or adding dice moves the whole table. How this style of mechanic compares with a single roll against a target is covered in the article on dice pools versus single rolls, so it is only introduced here.
Averages, spread and what “expected” means
The average result of a fair six-sided die is 3.5, the midpoint of 1 through 6. Averages add, so two dice average 7 and three dice average 10.5. An average of this kind is an example of expected value: what you would get per roll in the long run, not what any single roll gives. A single roll can never produce 3.5.
Spread matters too. A twenty-sided die is flat: every result from 1 to 20 is equally likely, so extreme results are common. Three six-sided dice have a similar average but a narrower, bell-like spread, so results near 10 or 11 occur far more often than 3 or 18. A designer picks between these on purpose. A flat die feels swingy and dramatic, while a sum of several dice feels steadier. The size of the typical swing is measured by standard deviation, which grows more slowly than the average as dice are added.
Modifiers, advantage and re-rolls
Games adjust the basic roll in a few common ways:
- Flat modifiers. Adding +2 to a twenty-sided roll against a target shifts the success chance by 2/20, or 10 percentage points, anywhere that the target is within the range of the die.
- Taking the better or worse of two rolls. This changes the shape of the odds rather than shifting them evenly. The effect is largest when success is near a coin flip, as explained under roll advantage.
- Re-rolling on top results. Some systems let a maximum roll trigger another die. These open-ended rules create long tails, as covered in exploding dice.
Each of these changes the same underlying table of outcomes. If you can list the equally likely outcomes and count the favourable ones, you can work out any of them by hand.
Checking your own dice
All of the above assumes fair dice. Real dice are physical objects, and cheap or worn ones can be slightly uneven. A repeated-roll test can show whether a die behaves roughly as expected, though small samples are always noisy. The sample size needed for a meaningful result is larger than most people expect, which is why the home test for dice fairness uses a simple comparison rather than a gut feeling.
A note on real-money dice games
Dice games played for money, such as craps, are built so that the organiser has an advantage. Understanding the probabilities explains why that advantage exists; it does not remove it. Games with a house edge lose money for players over time, outcomes in the short run are uncertain, and past rolls do not change future ones. If gambling is causing you harm, a local support service can help.
Using the odds in practice
For tabletop play, the useful habit is simple: list the equally likely outcomes, count the ones that matter, and divide. Check that your probabilities add to one. When a calculation gets long, switch to the complement or to a small simulation. A spreadsheet simulation of a few thousand rolls will usually land close to the exact answer, and it is a good way to confirm a hand calculation before you rely on it in a game design.