Lanceur de dés

How a roll is computed

Pick a die type — d4, d6, d8, d10, d12, d20, or d100 — a number of dice, and an optional flat modifier, and the tool rolls each die independently using crypto.getRandomValues() with rejection sampling, exactly the same underlying method used across RandoKit. Each individual die shows its own result, and the tool adds them together along with your modifier to produce the total — so "3d6+2" rolls three separate six-sided dice, sums them, and adds 2 to get the final number you'd read out at the table.

Sum of dice vs. a flat random range

This is the most misunderstood part of dice mechanics: rolling 2d6 is not the same as drawing one random number from 2 to 12. A single d12-style draw from 2-12 is uniform — every value from 2 through 12 has the same 1-in-11 chance. But 2d6 is a sum of two independent dice, and sums cluster toward the middle: there's only one way to roll a 2 (1+1) or a 12 (6+6), but six ways to roll a 7 (1+6, 2+5, 3+4, 4+3, 5+2, 6+1). Out of the 36 equally likely combinations of two six-sided dice, 7 comes up in 6 of them — about 16.7% of rolls — while 2 and 12 each come up in just 1 of 36, about 2.8%. The more dice you add to a sum, the more pronounced this bell-shaped clustering becomes, whereas a single flat draw from the same total range stays perfectly uniform across every value. If you want that flat, uniform behavior instead of a sum's bell curve, use the random number generator with min and max set to your range.

Worked example: 4d6 for an ability score

A common tabletop RPG method for generating a character stat is "roll 4d6, drop the lowest die, sum the rest." Roll four six-sided dice — say they come up 2, 4, 5, 6 — drop the lowest (the 2), and sum the remaining three: 4 + 5 + 6 = 15. Because you're dropping the worst of four independent rolls, this method skews results higher than a plain 3d6 roll would, which is exactly why it's favored for generating above-average starting stats. If your dice roller doesn't support "drop lowest" natively, roll 4d6 individually and do the drop-and-sum step by hand from the four numbers it shows you.

Advantage and disadvantage rolls

Some games use "advantage" (roll twice, keep the higher) or "disadvantage" (roll twice, keep the lower) instead of adding dice together. To simulate this, roll 2 of the same die type (commonly a d20) and manually take the higher or lower of the two shown results rather than their sum — the tool shows each die's individual value, so this works cleanly even without a dedicated advantage mode. Note that this is a different distribution from a flat single d20 roll: advantage pushes the expected outcome noticeably higher than 10.5 (the average of a plain d20), while disadvantage pushes it lower.

When to use this vs. other RandoKit tools

Use the dice roller whenever the shape of the randomness matters — tabletop games, board game rules that call for specific dice, or any situation modeled on physical dice mechanics. If you just need "a random number between X and Y" with every value equally likely, the random number generator is the more accurate tool, since it won't introduce the sum-of-dice clustering described above. For choosing among named options rather than numeric outcomes, see the wheel of names. And for more on the cryptographic randomness behind every roll, read how random number generators work.

Common mistakes

• Treating 2d6 or 3d6 like a uniform range when estimating odds — the middle values are genuinely far more likely, which matters for damage estimates and encounter balance.
• Forgetting to apply the modifier before comparing a roll to a target number, especially with stacked bonuses from multiple sources.
• Using a d100 roll when you actually need a flat 1-100 uniform draw for something unrelated to percentile tables — both give the same uniform distribution here, but conceptually a d100 is meant to represent percentile chance tables, not general-purpose number picking.

Fairness note

Each individual die face has an equal probability of appearing on any single die, drawn from the browser's cryptographic random source with rejection sampling to avoid modulo bias — the same guarantee described in the guide on Math.random() vs crypto.getRandomValues(). What isn't uniform, and isn't meant to be, is the sum across multiple dice — that clustering is a real mathematical property of adding independent random values, not a flaw in how the dice are generated.

Privacy

Every roll is generated locally in your browser, with no dice configuration or results sent to a server. Roll history stays in local storage on your device, and you can export results as .txt or .csv, or share a specific roll setup via an encoded URL, without any data leaving your device.