I'm asking more as an outside observer, and not with anyone who has any grievance with how games are played, nor am I judging or criticizing anything.

I've just always wondered, why use all the different types of dice? Why a 4 sided die for one thing, three 6 sided dice for another, and a twenty sided dice for something else? Couldn't you use, just for example, two 10 sided dice to represent a percentage between 1 and 100 (one die is tens and the other is ones, and they're differentiated by colour), and then use that for all instances?

If you needed to decide something was a one in four possibility, you just declare it a 25% chance, and roll the two tens. No matter what kind of possibility you wanted that are currently represented by various combinations of dice, it seems to me you could represent it within a range of one to one hundred.

Sure, a bag of all different shaped dice looks kind of cool, but is there a reason for them that goes beyond aesthetics? With just two 10 sided dice (or three if you wanted really fine gradients of percentage) wouldn't that make things simpler and more consistent without losing any of the variety of chances taken?

I thought maybe it had to do with probabilistic outcomes. For instance, if you roll two 6 sided dice, I believe that you're more likely to come up with some combination that adds up to 7 than you are with a combination that adds up to 2 or 12. So there's a bell curve of possible outcomes. Nonetheless, maybe I'm just not well versed enough in probabilities enough to see a difference, but I still think you could get the same equivalent likelihoods of an outcome by using a straight percentage, adjusting up or down depending on what you needed.

Again, I'm not trying to be contrarian, I'm just curious. Is there an objective reason for all the different dice?


closed as primarily opinion-based by SevenSidedDie, edgerunner, Brian Ballsun-Stanton Aug 3 '14 at 10:21

Many good questions generate some degree of opinion based on expert experience, but answers to this question will tend to be almost entirely based on opinions, rather than facts, references, or specific expertise. If this question can be reworded to fit the rules in the help center, please edit the question.

  • \$\begingroup\$ As a matter of fact, many, many games don't even use that many dice. Apocalypse World rolls everything with two d6's, and rolling the percentile 2d10 is even more common. \$\endgroup\$ – kviiri Aug 3 '14 at 8:22
  • 2
    \$\begingroup\$ This seems like a better question for statistics, as they can answer the statistical differences between dice rolling systems. Here it seems entirely too opinion-based. \$\endgroup\$ – Brian Ballsun-Stanton Aug 3 '14 at 10:32
  • 2
    \$\begingroup\$ Although this does seem a bit too subjective for good Stack Exchange style answers, feel free to join the Role-playing Games Chat for more free-ranging discussion on the topic. \$\endgroup\$ – BESW Aug 3 '14 at 11:05
  • 1
    \$\begingroup\$ I'm not particularly bothered that this question is on hold. If it's not for this site, then cool. But, I do feel that it's not subjective. The whole point is to ask if there is an objective reason. If someone could answer why various combinations of dice provide specifically different probabilities or effects, then wouldn't that be an objective discussion? \$\endgroup\$ – Questioner Aug 4 '14 at 2:22
  • 2
    \$\begingroup\$ @DaveMG The thing is, not all RPGs (and RPG players) use dice for the same reasons. Some pick their dice carefully for mathematical effect; others choose them based on novelty vs ease of access (d6s are easier to get than Fudge dice, but thus less cool); some just appeal to the desire to roll gigantic fistfuls of dice. There are literally thousands of RPG systems, and even more motives for how a group chooses which to use, so the idea that "why all the dice" can be objectively answered is going to require some uphill work to justify. \$\endgroup\$ – BESW Aug 4 '14 at 8:12

To make life easier.

Yes, you could reproduce every other probability curve - to within 1% accuracy, which is good enough for gaming - by rolling two ten-sided dice and taking a percentage, then looking up a conversion table for the outcome.

But which is easier and more natural...

  1. Roll two ten-sided dice in different colours. Convert them into a percentage. Pull out your percentage conversion table. Look up that, for example, a percentage roll of 13 means you rolled a four (but 17 would have been a five).

  2. Roll two six-sided dice and add them.

And that was the easy case.

World of Darkness games, for example, use multiple ten-sided dice - from one to ten - where greater skill means you get to roll more dice. So at a minimum, you'd need ten different probability conversion tables, one for each level of skill, listing the percentage chances of failure. Except it's worse than that, because the difficulty of the roll is variable, meaning that the number of d10 needing to show a high value to win changes... so to produce the same effect, you need about 80 different percentage tables. Just finding the right one would take a while!

And then there's the conversion back. D&D, for example, uses a d4 for the damage from small weapons - from one to four points of damage done. You can roll that with percentage dice - 25% is 1, 26%-50% is 2 - as you suggest. But you'll still need to convert it to a number from 1 to 4 when you're done. It's faster to just use a number from 1 to 4 in the first place.

Most die roll mechanics are, in fact, the simplest, easiest way to produce the desired random result - otherwise they get simplified until they are.

(Evil hat recently ran a Kickstarter to produce a FATE card deck for diceless resolution... but reproducing the probabilities of those four dice needs an entire deck of cards. And that's using just four three-sided dice; FATE has the simplest dice imaginable.)

We don't need the other dice. But they do make it much less work to open up a range of game design options.

  • 5
    \$\begingroup\$ 1% resolution isn't always good enough for gaming - the chance of rolling 3 or 18 on 3d6 is slightly under half a percent. (1 in 216, to be exact.) \$\endgroup\$ – Dave Sherohman Aug 3 '14 at 9:51
  • \$\begingroup\$ @DaveSherohman, you could use three 10 sided for percentages down to one decimal. Still less dice than the average bag'o'dice. \$\endgroup\$ – Questioner Aug 4 '14 at 2:23
  • \$\begingroup\$ Tynam, thank you for this response. You mention something that others have brought up, which is the need to do some kind of "conversion" of percentages to probababilities. I don't see that as a significant issue, though, because those conversions should be considered and created at the time of making rules, not during individual adventures. Perhaps my question wasn't clear. I wasn't suggesting that it would be efficien to take a game that currently uses multiple dice and emulate it with 2D10, but that an equally variable but more efficient game system could be created. \$\endgroup\$ – Questioner Aug 4 '14 at 2:28
  • \$\begingroup\$ @DaveMG The issue that then comes up is matching the scale of other numbers that add to or otherwise interact with the roll. A +1 modifier means a different amount of change to a roll of 1d4 than it does to a 1d20; Sure, you could convert them to percentages, but adding 25 or 5 to a roll is (very slightly) less convenient than adding 1. \$\endgroup\$ – GMJoe Aug 4 '14 at 4:18
  • 1
    \$\begingroup\$ @DaveSherohman: I chose those words carefully. I agree that almost all games use dice conventions that are not represented accurately at 1% resolution. (Even 2d6 isn't.) But I would argue very strongly that no RPG exists where gameplay is harmed by a less-than-1% probability difference. I said that 1% is good enough for gaming purposes. (Is AD&D actually a noticeably different game if your probability of an 18 stat in character creation becomes 1% instead of 0.463%? No. And that's the biggest difference it would make.) \$\endgroup\$ – Tynam Aug 5 '14 at 13:55

Short answer: Yes, your hunch is right: because probability curves, and easy access to them.

Sure, you could calculate (for example by using AnyDice) in advance the percentage chances of rolling a 2, a 3, a 4 etc on a 3d6, but mapping and rolling everything so would get messy and totally inconvenient practically in an instant. You want predictably varying distribution, bell curves, etc, and all this at your fingertips without having to use a calculator or other app.

Also, dice are nice. It's great to roll them. Look, how they twist and spin and decide your fate. :D


There's quite a few reason why many different kind of dice can be desirable, I'll concentrate on the mathematical reasons.

First off, in games like D&D most of the different dice are used to calculate damage. A greataxe deals a base 1d12 of damage while a dagger deals 1d4. Using different dice to model those different weapons is much easier than using a giant dice and dividing the result by some integer (especially since 100 isn't divisible by 12.)

The other main reason is that different kind of dice allow for different probability distributions. If you use a d20 to try to jump across a chasm in D&D You are just as likely to get a 1, 10 or 20 on the die. However, if you used 3d6 you would get a result between 3 and 18, but not every result would be as likely.

A 3 and an 18 would only happen once in every 216 rolls while 9-12 would happen about 10% of the time each. This may seem subtle, but that kind of distribution makes exceptional failure and success less frequent and as such more interesting. Small bonuses to those rolls also become more noticeable since most of your results are very close together.

In the end those could be simulated with some kind of generic dice, but tabletop RPGs require a lot of preparation, memorization and concentration to play and you really don't want your players to have to look up some kind of result table even just once in a while as that would slow the game down unnecessarily.


Aside from the mathematical reasons, using different types of dice allows for things such as rolling to hit and damage simultaneously rather than sequentially - most groups, in my experience, find it faster to throw a d20 and a d8 on the table at the same time than to roll the d20, decide if it's a hit, and only then roll the d8.

About a year ago, some bloggers took this to the extreme and started posting "roll all the dice tables", which were sets of (usually) six tables, one each for d4, d6, d8, d10, d12, and d20, to determine different aspects of a single thing with the intention that you can grab one die of each type and roll all of them at once instead of rolling on each table in sequence. For example, On A Roll All the Dice Table: Keys


Not all modern die rolling systems directly translate to an exact percentage chance. For instance, any open-ended or exploding die system can generate an infinitely-large result a small percentage of the time. Or, systems that use a degree of success system.

  • 2
    \$\begingroup\$ Degree of success is quite possible in a percentile system. \$\endgroup\$ – SevenSidedDie Aug 3 '14 at 8:51

Not the answer you're looking for? Browse other questions tagged or ask your own question.