Single die
Roll a d8
A 8-sided die, rolled in your browser. Every face has an equal 12.5% chance, which works out to 1 in 8.
The odds on a d8
A single die is flat: no total is more likely than any other. The curve only appears once you roll more than one, which is what the 2d8 page is for.
- Lowest
- 1
- Highest
- 8
- Average
- 4.50
- Each face
- 12.5%
- Odds
- 1 in 8
Every bar the same height. That flatness is the whole point of a single die: 12.5% per face, no favourites.
What a d8 is for
Longsword and rapier damage, and hit dice for the middleweight classes.
Hitting a target number
What a flat d8 beats, and how much two-dice advantage moves it.
| Need | Flat | Advantage | Disadvantage |
|---|---|---|---|
| 2+ | 87.5% | 98.4% | 76.6% |
| 3+ | 75.0% | 93.8% | 56.2% |
| 4+ | 62.5% | 85.9% | 39.1% |
| 5+ | 50.0% | 75.0% | 25.0% |
| 6+ | 37.5% | 60.9% | 14.1% |
| 7+ | 25.0% | 43.8% | 6.25% |
| 8+ | 12.5% | 23.4% | 1.56% |
How to use this roller
Tap the die or press the roll button. To roll several at once, change How many. To add a bonus, put it in Modifier: entering 3 turns a d8 into 1d8+3. Negative numbers work for penalties. The keyboard shortcut is R.
Nothing is stored on a server and no account is needed. The page works offline once loaded, which matters if the venue has poor signal.
Common questions about the d8
What are the odds of rolling a 8 on a d8?
Exactly 12.5%, or 1 in 8. Every face on a fair d8 is equally likely, so a 8 is no rarer than a 7.
What is the average roll on a d8?
4.5. That is the midpoint of 1 and 8, and it is worth knowing because the average of a die is always (sides + 1) divided by 2. Over many rolls your results converge on it, but any single roll tells you nothing.
Does rolling with advantage change the odds of a 8?
Advantage is a d20 mechanic, but the maths applies to any die. Rolling two d8 and keeping the higher would raise your chance of a 8 from 12.5% to 23.4%.
Is this d8 roller actually random?
Every roll comes from the Web Crypto API rather than Math.random, and uses rejection sampling to discard values that would skew the result. A plain modulo of a random number is biased when the range does not divide evenly, which for a d8 would quietly favour the low faces.