Same average, wildly different distributions. The math behind why game designers choose one over the other — and what it means for your character build.

Published: 2026-09-16By: Toolhub Admin · About the author

Roll 2d6 a thousand times and you'll get 7 more than any other number. Roll 1d12 a thousand times and every number from 1 to 12 shows up roughly equally. Both dice setups cover a similar range. Both average out to similar values. But the shape of the probability distribution is completely different — and that shape is what actually determines how the dice feel at the table.

This isn't trivia. The choice between multi-die and single-die mechanics is one of the most consequential design decisions in tabletop games. It affects how often extreme results occur, how reliable damage output is, and whether a player's strategy should lean toward consistency or swings. Here's the math.

The distributions side by side

1d12 is a uniform distribution. Each face (1 through 12) has exactly 1/12 probability — about 8.33%. Rolling a 1 is exactly as likely as rolling a 7 or a 12. There's no "most likely" result.

2d6 is a triangular distribution (discrete version of a bell curve). There are 36 possible outcomes (6 x 6), and they cluster around the middle:

Sum   Ways to roll it   Probability
 2    1  (1+1)           2.78%
 3    2  (1+2, 2+1)      5.56%
 4    3                   8.33%
 5    4                  11.11%
 6    5                  13.89%
 7    6                  16.67%
 8    5                  13.89%
 9    4                  11.11%
10    3                   8.33%
11    2                   5.56%
12    1                   2.78%

Rolling a 7 on 2d6 is six times more likely than rolling a 2 or 12. The extremes are rare; the middle is crowded. This is the Central Limit Theorem in miniature — adding independent random variables pushes the sum toward a normal distribution, and even two dice are enough to see the effect clearly.

Use a dice roller to simulate a few hundred rolls of each and watch the histogram form. The 1d12 histogram stays flat. The 2d6 histogram peaks sharply at 7.

Variance and what it means at the table

The average (expected value) for 1d12 is 6.5. For 2d6 it's 7. Close enough that the averages barely matter for comparison. What matters is the spread.

         Mean    Variance    Std Dev
1d12     6.50    11.917      3.452
2d6      7.00     5.833      2.415

1d12 has roughly double the variance of 2d6. The standard deviation — the practical measure of "how far from average will a typical roll be" — is 3.45 for 1d12 versus 2.42 for 2d6. In concrete terms: with 2d6, about 68% of your rolls will fall between 5 and 9. With 1d12, the same 68% band stretches from 3 to 10.

Higher variance means more swings. More dramatic highs, more painful lows, less time near the average. Lower variance means more clustering around the mean — predictable, reliable, fewer surprises.

D&D's greatsword vs greataxe

The classic example. In D&D 5th Edition, the greatsword deals 2d6 slashing damage and the greataxe deals 1d12 slashing damage. Both are heavy, two-handed martial weapons. The choice between them is a choice between the distributions above.

Greatsword (2d6): Average damage 7. Minimum 2, maximum 12. You'll rarely roll below 4 or above 10. Consistent, reliable damage output. Over 10 rounds, your total damage will cluster tightly around 70.

Greataxe (1d12): Average damage 6.5. Minimum 1, maximum 12. You'll hit every value from 1 to 12 with equal frequency. Over 10 rounds, your total could plausibly be anywhere from 35 to 95. More volatile.

The greatsword wins on average (7 vs 6.5) and on consistency. The greataxe's only statistical advantage is hitting 12 — it has an 8.33% chance versus the greatsword's 2.78%. If your character has features that trigger on maximum damage or that benefit from high single-roll results (like the Barbarian's Brutal Critical, which adds extra dice on a crit), the greataxe's flat distribution makes those features more impactful.

Check the expected damage of your own weapon setup with a damage calculator — factor in your modifier, advantage/disadvantage, and crit features to see which weapon actually deals more over a full combat.

Expected damage per round with modifiers

Raw dice are only part of the damage formula. In D&D, you add your ability modifier (typically +3 to +5) to every hit. That flat bonus doesn't change the variance of the dice roll, but it does change the coefficient of variation — the ratio of standard deviation to mean. With a +5 modifier:

Greatsword: 2d6+5 = avg 12, std dev 2.42 (CV = 20.2%)
Greataxe:   1d12+5 = avg 11.5, std dev 3.45 (CV = 30.0%)

The modifier compresses the relative volatility for both weapons, but the greataxe is still 50% more volatile in relative terms. At lower levels with a +3 modifier, the gap is even more pronounced. As modifiers increase, both weapons converge toward "the modifier is doing most of the work and the dice matter less" — which is one reason high-level D&D combat can feel less exciting than low-level.

How game designers use this

The uniform-vs-bell-curve choice appears everywhere in game design, not just D&D weapons:

Reliability mechanics. Games that want skill to dominate luck use multi-dice systems. GURPS uses 3d6 for skill checks — an extremely tight bell curve (mean 10.5, std dev 2.96 on a range of 3-18) where moderate skills succeed consistently and extreme results are genuinely rare. A skill of 12 succeeds 74% of the time. Raising it to 14 succeeds 91%. The bell curve rewards incremental investment.

Swingy mechanics. Games that want dramatic moments use single-die systems. D&D's d20 for attack rolls is uniform — a +1 bonus always improves your chances by exactly 5%, and a level-1 commoner has a real shot at hitting a dragon (just needs a natural 20). The flat distribution keeps uncertainty high and makes every roll feel consequential.

Damage vs hit resolution. Many games deliberately use different distributions for "did you hit?" (swingy, uniform, dramatic) versus "how much damage?" (consistent, multi-die, predictable). D&D does exactly this — d20 for attacks, multiple dice for damage. The emotional arc is: uncertain whether you'll hit (exciting), then reliable-ish damage when you do (satisfying).

Pool systems. Games like Shadowrun (roll Xd6, count dice showing 5+) and World of Darkness (roll Xd10, count successes) add another dimension. More dice in the pool means more successes on average with less relative variance — the Central Limit Theorem again. A character with 8 dice feels meaningfully more consistent than one with 3 dice, even beyond the raw average difference. The distribution shape changes with the pool size.

The takeaway for players and designers

If you want consistency — reliable results that cluster near the average with rare extremes — use more dice with fewer faces. 2d6 is more predictable than 1d12. 3d4 (range 3-12, mean 7.5, std dev 1.94) is more predictable still. The more dice you add, the tighter the bell curve.

If you want drama — genuine uncertainty where any single roll might be brilliant or catastrophic — use a single die. 1d12 swings harder than 2d6. 1d20 swings harder still. The uniform distribution treats every outcome as equally likely, and that equality is what creates the tension.

Same range, same average, completely different game feel. The distribution is the design.

← All articles