Dice Probability: The Odds Behind Every Roll

On a fair die, every face has the same chance of landing up. On a d6 that’s 1 in 6, about 16.7%. On a d20 it’s 1 in 20, or 5%. For two or more dice, you count the ways each total can happen and divide by every possible outcome. That’s the whole trick, and the rest of this guide is just that trick wearing different hats (and occasionally a tiny wizard robe).

We sell tabletop dice for a living, so we have strong feelings about the odds. Mostly: they’re not out to get you. Your d20 just has a sense of humor.

TL;DR

• One fair die: probability of any face = 1 divided by the number of sides. A d6 gives 1/6 (about 16.7%), a d20 gives 1/20 (5%).
• Two d6: 36 possible outcomes, and 7 is the most common total because 6 of them make it (1/6, or 16.7%).
• Average roll of a die = (sides + 1) / 2. A d20 averages 10.5, a d8 averages 4.5.
• Advantage on a d20 takes your odds of a natural 20 from 5% to 9.75%. Not magic, just arithmetic.
• Probability assumes a fair die. Balance matters, and you can check it at home.

Table of Contents

• The one-die rule: favorable over total
• Two dice: why 7 always shows up to the party
• The d20 math every player eventually asks about
• Probability assumes a fair die
• Why a D&D set has so many shapes
• Our take: the math is cold, the table is warm
• Picking dice that deserve your math
• Worked examples to try yourself
• Do digital rollers change anything?
• FAQ

The one-die rule: favorable over total

Probability, in the classic sense, is simple. Count the outcomes you want, count all the equally likely outcomes, and divide. OpenStax defines it as the number of elements in your event divided by the number of elements in the sample space, as long as every outcome is equally likely.

For a d6, the sample space is {1, 2, 3, 4, 5, 6}. Want a 3? One favorable outcome out of six, so the probability is 1/6. The Mathematics LibreTexts odds chapter walks through exactly this case, and also points out that the odds in favor of a 3 are 1:5, because one outcome is a 3 and five are not. Probability and odds are cousins, not twins: 1/6 versus 1 to 5.

A few handy facts that hold for any die:

• Probabilities are always between 0 and 1 (or 0% and 100%). You can write them as a fraction, a decimal, or a percentage, as Peter Dunn’s text notes.
• All the faces together add up to 1. Six faces at 1/6 each: 100%. The universe is balanced.
• Every roll is independent. The die has no memory, no grudge, and no idea you’ve missed three attacks in a row.

Here’s the quick reference for the standard polyhedral set:

The average comes from (sides + 1) / 2. Add the lowest and highest face, halve it, done. A d8 is (8 + 1) / 2 = 4.5, which is a number you can never actually roll. Dice are weird like that.

Two dice: why 7 always shows up to the party

Roll one die and every number is equally likely. Roll two and suddenly the totals are not. This is the part that trips people up, and it’s the reason “why is 7 the most likely roll?” gets typed into search bars all day.

Two six-sided dice make 36 equally likely combinations (6 × 6). The totals are not equally likely because some totals can be built in more ways. A 7 can be 1+6, 2+5, 3+4, 4+3, 5+2, or 6+1. That’s six ways. A 2 can only be 1+1. Michael Shackleford’s dice tables lay out these counts, and the two-dice row matches what we’ve got here.

See the shape? It’s a tent, peaking at 7. The distance from 7 tells you how rare a total is. Snake eyes and boxcars are each a 1-in-36 event, which sounds rare until you roll dice for a few hours and watch it happen at the worst possible moment.

The “sum the previous column” trick

Need three dice? Don’t list 216 combinations by hand. The Shackleford tables build each new dice count from the one before: the number of ways to hit a total with n dice is the sum of the ways to hit the nearby totals with n−1 dice (one term for each face of the new die). It’s tedious by hand but perfect for a spreadsheet.

The bigger the pile of dice, the more the tent smooths into a bell curve. That’s why 8d6 for a fireball clusters around 28 (8 × 3.5, since averages simply add up) and almost never gives you the extreme low or high end. Your DM is rolling a lot of dice partly so the average behaves.

The d20 math every player eventually asks about

Most tabletop RPGs lean on the d20, so here’s where probability earns its keep at the table.

Hitting a target number. Each point on the d20 is worth 5 percentage points. To beat a DC of 15 you need 15 or higher: that’s 6 faces (15 through 20), or 30%. Beat a DC of 11 or higher? Ten faces, 50%. With a +5 modifier you need a 10 or higher: 11 faces, 55%.

Natural 20. One face in twenty: 5%. Natural 1: also 5%. Over a long campaign, they come up equally often. Your memory just files the 1s under “personal betrayals.”

Advantage. You roll two d20s and take the higher. To fail you need both to be low, and independent events multiply. Radford University’s tutorial calls this the product rule: for independent events, multiply the individual probabilities. So:

• Chance of not rolling a 20 on one die: 19/20 = 0.95
• Chance neither die is a 20: 0.95 × 0.95 = 0.9025
• Chance at least one is a 20: 1 − 0.9025 = 9.75%

And for a 50% roll (needing 11+), advantage gives 1 − (0.5 × 0.5) = 75%. Disadvantage flips that: you need both dice to succeed, so 0.5 × 0.5 = 25%. Advantage is worth the most when the roll is near a coin flip, and a lot less when you’re already almost sure to succeed.

“At least one” problems: flip them around

Questions like “what’s the chance of at least one 6 in four rolls?” are easier backwards. Work out the chance of no 6s, then subtract from 1. Chance of no 6 on one d6 is 5/6. Four dice: (5/6)⁴ ≈ 0.482. So at least one 6 shows up about 51.8% of the time. Slightly better than a coin flip, and a lot worse than most people guess.

One warning from that same Radford tutorial: when events overlap (they aren’t mutually exclusive), you can’t just add their probabilities. You’ll double-count the overlap. “Roll an even number or a number above 4” on a d6 is {2, 4, 6} plus {5, 6}, and the 6 only counts once: 4 faces out of 6, not 5.

Quick one: adding modifiers

Flat bonuses just slide the whole distribution up. A +3 on 2d6 doesn’t change the shape, it changes the average from 7 to 10. Rolling 2d6+3 and needing 12 or more means you need a base total of 9 or more: 4 + 3 + 2 + 1 = 10 ways out of 36, or 27.8%.

Probability assumes a fair die

Everything above has a quiet footnote: equally likely outcomes. Probability math is only as honest as the die. If one face is slightly heavier, or the edges are rounded unevenly, or there’s a bubble inside the plastic, the 1/20 per face is a polite fiction.

This is why dice makers and gamers talk about balance. If you want to go down that rabbit hole, we’ve written a guide on how to choose balanced gaming dice and another on what dice balance actually is and how to check it. The short version: you can test a die by rolling it a few hundred times and seeing if any face is getting unreasonable favoritism. If your d20 throws a 17 every third roll, it may not be cursed, it may just be a bad die.

The same logic works in reverse for games that use deliberately uneven dice. If you’re curious how probability changes when the rules of equal likelihood bend, chess played with a die is a fun example of how one random roll reshapes strategy, because you have to think in odds rather than certainties.

Do different materials change the odds?

Not in theory. A fair metal d20 and a fair resin d20 have the same 5% per face. In practice, material affects how a die tumbles and how much it weighs, which is why folks ask why metal dice are so heavy. Weight alone doesn’t skew results. Poorly made geometry can.

Why a D&D set has so many shapes

Different dice give different curves. A d20 is flat: every outcome is equally likely, which makes it swingy. A pile of d6s clumps around the middle, which makes damage predictable. That’s deliberate design, and we dug into it in our pieces on why D&D uses so many different dice and the function of each die in the set. If notation like 2d6+3 still looks like a license plate, here’s a brief guide to dice notation and names.

Our take: the math is cold, the table is warm

We get asked a version of this a lot: “Do my dice hate me?” Here’s our honest answer, as people who spend an unreasonable amount of time around dice.

Probability tells you what happens across lots of rolls. It says nothing about the next one. Three natural 1s in a row is a story, not evidence. Streaks happen all the time in random sequences, and the stories we remember are the painful ones. Rolling a 20 when nothing’s on the line doesn’t get told at the table; failing a save against the final boss does.

What we do care about is whether a die is well made. A fair die lets the math hold, and a beautiful one makes the table care. That’s why we make our sets in liquid core, gemstone, metal, resin and wood: the odds are the same, but the experience of rolling the thing is not. Knowing the math makes you braver at the table too. You’ll know when the long shot is worth taking and when the DC 18 is a bad idea, and you’ll get to be dramatic about it either way.

Probability is a tool for fun, not a curse on your character sheet. Roll the dice, enjoy the chaos, and blame the bard.

— Teresa

Picking dice that deserve your math

If you’ve read this far, you’re officially a person who calculates the odds of a natural 20 for fun. Congratulations. Now you need dice worthy of that knowledge.

Our metal, resin and wood dice cover the classic sets. If you like a little spectacle, the liquid core collection has dice with something swirling inside, and our Gemstone Series leans toward the look of polished stone. Can’t choose? The Random Cache is our mystery bundle, which is a nice way to practice accepting random outcomes (the probability of liking what you get is, we hope, high). Our guarantee is an unconditional money-back policy, so the only risk on the table is your next saving throw.

Planning a whole campaign? Our guides on how many gaming dice a DM needs and the best dice for tabletop RPG campaigns will help you decide what to put in your bag.

Worked examples to try yourself

Flex the muscles with a few problems. Answers follow, no peeking (okay, peek a little).

1. What’s the probability of rolling at least an 8 on 2d6?
2. What’s the chance of rolling a 1 or a 2 on a d8?
3. You roll three d6. How likely is a total of 10 or 11?
4. A d20 attack needs a 14 or higher. What’s the chance with advantage?

Answers.

1. Ways to make 8 through 12: 5 + 4 + 3 + 2 + 1 = 15 of 36, so 41.7%.
2. Two favorable faces out of 8: 25%.
3. Three dice have 216 combinations. Totals of 10 and 11 each have 27 ways, so 54/216 = 25%. (A total of 10 or 11 is the most common outcome for 3d6, because 10.5 is the average.)
4. One die succeeds on 14 through 20, which is 7 faces, so 35%. Both fail at 65%: 0.65 × 0.65 = 0.4225. Success with advantage is 1 − 0.4225 = 57.75%.

A good habit: before you reach for a calculator, ask whether you’re counting a single die (divide by sides), a sum (count combinations), or an “at least one” (flip it and subtract from 1). Nearly every dice question falls into one of those three buckets.

Do digital rollers change anything?

A well-built roller uses a random number generator that treats every face as equally likely, so the probability is the same as a fair physical die. Whether you prefer virtual or real is a vibes question, and we’ve covered digital dice rollers if you want the pros and cons. We’ll admit we’re biased: nothing in an app clatters across a tabletop and interrupts the bard mid-solo.

FAQ

What is the probability of rolling a 3 on a six-sided die?

It’s 1/6, which is about 0.1667 or 16.67%. There’s one favorable face out of six equally likely faces, assuming the die is fair. The odds in favor are 1 to 5, since five faces aren’t a 3.

Why is 7 the most likely roll with two dice?

Because more combinations make 7 than any other total. Six of the 36 possible outcomes add to 7 (1+6, 2+5, 3+4, 4+3, 5+2, 6+1), so its probability is 1/6. A 2 or a 12 has only one combination each.

How do I calculate the probability of a sum on multiple dice?

Count how many combinations produce your total, then divide by the total number of outcomes (6 to the power of the number of d6s). For two dice that’s 36, for three it’s 216. Tables like these give the counts for many dice at once, or you can build them one die at a time in a spreadsheet.

What is the average roll of a d8?

It’s 4.5, from (8 + 1) / 2. The same formula gives 3.5 for a d6, 6.5 for a d12 and 10.5 for a d20. For several dice, add their averages and then add any flat modifier.

Does a heavier or fancier die change the probability?

Not if the die is fair. Material and style change feel, weight and looks, but each face still lands with the same chance. What changes the odds is poor balance or sloppy geometry, which is worth checking before you trust a die with your campaign finale.

Recommended

• Dice Balance: What is it?
• A Brief Guide to Dice Notation and Names
• How to Choose Balanced Gaming Dice: A D&D Guide
• Best Dice for Tabletop RPG Campaigns: 2026 Guide