Dice Odds Explained: Averages, Probabilities, and Reading Dice Notation

What dice notation means

Every dice roll in a tabletop game is written in the same compact shorthand: NdX. The first number is how many dice you throw, the letter d stands for "die", and the last number is how many sides each one has. So 1d20 means "roll one twenty-sided die", 2d6 means "roll two six-sided dice and add them together", and 4d8 means "roll four eight-sided dice and sum the results". A trailing number tacks on a fixed modifier — 1d20+5 rolls a d20 and adds five to whatever it shows. That is the entire language, and once you can read it you can describe any roll in any game. The dice roller on this page takes those three pieces — how many dice, which die, and the modifier — and rolls them for you, showing each die separately as well as the total.

The seven standard polyhedral dice are the d4, d6, d8, d10, d12, d20, and d100. The d6 is the familiar cube from board games; the d20 is the icosahedron at the heart of Dungeons & Dragons; and the d100, or percentile die, produces a number from 1 to 100 and is usually rolled as two d10s — one read as tens, one as units. Different games lean on different dice, but the notation and the underlying maths are identical across all of them.

How dice probability works

A single fair die is the simplest random object in probability: a discrete uniform distribution. Every face is equally likely, so a d6 lands on each of 1 through 6 with a 1-in-6 chance, and a d20 lands on each of 1 through 20 with a 1-in-20 chance — a flat 5% per face. Two numbers describe the whole picture. The average (or expected value) of a die with X sides is (X + 1) / 2, which is 3.5 for a d6 and 10.5 for a d20. The spread, measured as the standard deviation, is √((X² − 1) / 12) — about 1.71 for a d6 and 5.77 for a d20.

The moment you roll more than one die, the picture changes in a way that trips a lot of people up. When you add dice together, the averages simply add: two d6 average 3.5 + 3.5 = 7, and three d6 average 10.5. But the shape of the results stops being flat. With a single die every outcome is equally likely; with a sum of dice the middle values become far more common than the extremes, because there are many more ways to make a middle total than an extreme one. This clustering is the single most important idea in dice probability, and it is why a game designer's choice of "roll 3d6" versus "roll 1d20" completely changes how a game feels.

The formulas that govern a sum of N identical dice are worth keeping to hand:

Average of NdX   = N × (X + 1) / 2 Lowest possible  = N          (every die shows 1) Highest possible = N × X      (every die shows its top face) Number of ways to roll a total = the count of dice combinations that sum to it

Feed any combination into the dice roller and it applies these rules instantly, but understanding them is what lets you judge a roll before you make it. If you want to work the average of an arbitrary roll out for yourself, the expected value calculator handles any weighted set of outcomes, and the average calculator is handy for tallying a session's worth of results.

Worked example: the shape of 2d6

Two six-sided dice are the classic example because they show the clustering effect clearly. There are 6 × 6 = 36 equally likely combinations. A total of 7 can be made six ways (1-6, 2-5, 3-4, 4-3, 5-2, 6-1), so it comes up 6 / 36 ≈ 16.7% of the time. A total of 2, by contrast, can only be made one way (1-1), so it appears just 1 / 36 ≈ 2.8% of the time — and the same is true of 12 (6-6). The full distribution is a neat triangle:

Total   Ways   Probability 2       1        2.8% 3       2        5.6% 4       3        8.3% 5       4       11.1% 6       5       13.9% 7       6       16.7%   ← most likely 8       5       13.9% 9       4       11.1% 10       3        8.3% 11       2        5.6% 12       1        2.8%

Seven is not lucky — it is just the total with the most combinations behind it, sitting exactly at the average of 7. This is why craps and Monopoly feel the way they do: the middle totals dominate, and betting on the edges is a long shot. Roll 2d6 a few dozen times in the dice roller and you will watch the totals pile up around 6, 7, and 8 exactly as the table predicts. Add a third die and the pile grows tighter still: 3d6 averages 10.5 but lands on 10 or 11 about a quarter of the time, while the extremes of 3 and 18 each turn up less than once in 200 rolls.

Why "same average" does not mean "same die"

Here is the comparison every game designer thinks about. A 1d20 and a 3d6 both average 10.5. But a d20 is dead flat — every result from 1 to 20 is equally likely, so wild swings are routine and a 1 is exactly as common as a 10. Three d6 produce a range of 3 to 18 that bunches hard around the middle: results of 10 and 11 each appear 12.5% of the time, while a 3 or an 18 shows up under half a percent of the time. Same centre, utterly different feel. A d20 game is swingy and dramatic; a 3d6 game is steady and predictable, where a skilled character can count on a near-average result. Neither is "more random" — they simply allocate their randomness differently, and you can see it in the standard deviations: 5.77 for the d20 against 2.96 for 3d6.

How advantage and disadvantage change the odds

The advantage mechanic in modern D&D is the cleanest real-world lesson in dice probability there is. With advantage you roll two d20s and keep the higher; with disadvantage you roll two and keep the lower. It sounds like a small tweak, but the effect is large and, crucially, uneven.

Say you need an 11 or better on a d20 to succeed — a straight 50% chance. With advantage you succeed unless both dice fail, and the chance of two failures is 0.5 × 0.5 = 0.25, so your success rate jumps to 75% — a full 25 percentage points. That is the biggest boost advantage can give, and it lands precisely when the roll is a coin-flip. When you only need a 2 to succeed (a 95% chance already) advantage adds almost nothing, because you were going to make it anyway; when you need a natural 20 (5%), advantage only lifts you to 9.75%. The general rule: advantage helps most on the rolls that matter most, the tight ones around 50/50. Across the whole range it raises your average d20 result from 10.5 to about 13.8, and disadvantage drops it to about 7.2.

Common misconceptions

Thinking a number is "due"

Dice have no memory. After rolling four 1s in a row on a d6, the chance of a fifth 1 is still exactly 1 in 6 — the die does not know or care what it did before. Believing that a run of low rolls makes a high roll "due" is the gambler's fallacy, and it has cost people real money at real tables for centuries. Each roll is independent; the long-run average only emerges over hundreds of rolls, not as a correction applied to the next one.

Confusing the average with the most likely result

For a single die there is no "most likely" result — every face ties. For a sum of dice the most likely total does sit at the average (7 on 2d6), but the average is still just one outcome among many, and on 2d6 you miss it 83% of the time. Planning as though you will roll the average every time is a good way to be surprised.

Assuming more dice means a wider spread of results

It is the opposite. Rolling 10d6 gives a much larger range (10 to 60) than 1d6, but relative to that range the results cluster tightly around the average of 35 — extreme totals become vanishingly rare. Piling on dice makes the outcome more predictable, not less, which is exactly the central-limit effect that turns a heap of dice into a bell curve.

Are dice — physical and digital — actually fair?

A perfectly fair die gives every face an equal chance, and the maths on this page assumes exactly that. Real dice are close but not perfect. Cheap injection-moulded dice with rounded corners and deeply carved pips can be measurably biased, because material is removed unevenly and the shape is not quite symmetric. Casino dice are the gold standard: machined to tight tolerances with sharp edges and flush pips filled with paint of the same density as the body, so no face is favoured. For most tabletop play the bias in ordinary dice is far too small to notice over a session, but it is real, and it is why serious games use quality dice.

A digital roller like this one has no physical imperfections at all. It draws each result from your browser's built-in pseudorandom number generator (the standard Math.random function), which produces numbers that are uniformly distributed and, for any practical gaming purpose, unpredictable. It is not a cryptographic source — you would not secure anything with it — but for deciding whether your rogue sneaks past the guard it is more even-handed than any plastic die, and it never rolls off the table. If you want to sanity-check that a run of digital rolls really is uniform, tally the results and compare their spread against the theoretical figure using the standard deviation calculator.

When the maths runs out

Simple sums and single dice are fully covered by the formulas above, but some questions get genuinely intricate — the odds of rolling at least three 6s on eight dice, the distribution of "4d6, drop the lowest" for rolling up a character, or the chance of beating an opposed roll. Those involve counting combinations or working through a binomial distribution, and while the principles never change, the arithmetic is best handed to a tool. The binomial distribution calculator answers "how likely am I to get at least k successes in n independent tries", which covers a huge share of dice questions once you frame "rolling a 6" as a success with probability 1/6.

Frequently asked questions

What is the average roll of two dice? Two six-sided dice average 7, because each die averages 3.5 and the averages add. More generally the average of any NdX roll is N × (X + 1) / 2, so 3d6 averages 10.5 and 2d10 averages 11. The average is also the peak of the distribution for a sum of dice — 7 is both the mean of 2d6 and its single most common total.

Why is 7 the most common result on two dice? Because it has the most combinations behind it. There are six ways to make 7 (1-6, 2-5, 3-4, 4-3, 5-2, 6-1) out of 36 equally likely outcomes, while a 2 or a 12 has only one way each. More combinations means a higher chance, and 7 sits in the middle where the combinations are most plentiful.

How much does advantage actually help in D&D? It depends entirely on what you need to roll. Advantage is worth up to +25 percentage points when your target is around 50/50, shrinking towards nothing at the extremes where you were almost certain to succeed or fail anyway. On average it lifts a d20 roll from 10.5 to roughly 13.8 — an effective bonus of about +3 to +5 on the rolls that count.

What is the difference between rolling 3d6 and 1d20? Both average 10.5, but the spread is completely different. A d20 is flat — every number 1 to 20 is equally likely — so it is swingy and unpredictable. Three d6 span 3 to 18 and cluster around the middle, with 10 and 11 each appearing 12.5% of the time and the extremes almost never. Games use the flat d20 for drama and the bell-shaped 3d6 for consistency.

How do I roll ability scores for D&D? The traditional method is "4d6, drop the lowest": roll four six-sided dice, discard the smallest, and add the other three. This nudges results upward — the average score works out to about 12.24 rather than the 10.5 you would get from a plain 3d6 — and you repeat it six times, once for each ability. Rolling four dice and keeping the best three is easy to do by hand with the dice roller set to 4d6.

Are online dice rollers rigged? A well-built roller is not — it draws from a pseudorandom number generator that is uniformly distributed and unpredictable for gaming purposes, which actually makes it fairer than a mass-produced plastic die that may carry a small physical bias. It is not suitable for cryptography or real-money gambling, where a dedicated hardware or cryptographic source is required, but for tabletop play it is even-handed and convenient.

What are the odds of rolling at least one 6 on four dice? About 51.8%. It is easier to work out the chance of no sixes — (5/6)⁴ ≈ 48.2% — and subtract from 100%. This "one minus the chance of none" trick is the fastest way to answer any "at least one" dice question, and it is the same logic the binomial distribution calculator automates for larger numbers of dice.