Dice Roller

Dice notation, probability, and why a fair RNG matters

Tabletop games settle disputes with dice because a die is cheap, physical, and (mostly) impartial — but a physical die can be shaved, warped, or dropped on a slope, and rolling eight six-sided dice for a fireball by hand is slow. A dice roller solves the mechanical half of that problem: it produces the same range of outcomes as the physical object, at any quantity, with a random source that doesn't degrade the way a worn plastic die does. That matters for anything scored by chance — a Dungeons & Dragons attack roll, a Yahtzee turn, a board game that moves a token, or a classroom probability demonstration.

The shorthand that makes this practical is dice notation. It was popularized by tabletop RPGs in the 1970s as a compact way to write "roll some dice and add a number" inside a rulebook, and it has since become the default way games, apps, and even spreadsheet formulas describe a roll.

Reading and entering dice notation

1. Notation is written as count "d" sides, optionally followed by + or − and a flat modifier — no spaces, no other symbols.
2. The count can be left out when it is 1: d20 means the same thing as 1d20.
3. The modifier is added once to the total after every die has been summed, never to each individual die.
4. Type it into the notation field and press Roll; the notation box overrides whatever die buttons and quantity are selected, so you don't need to also set them.

Why the total isn't flat like a single die

A single die is uniform: every one of its faces has exactly the same chance, 1 in the number of sides. Add a second die and that stops being true, because the total is a sum, and sums have more ways to hit the middle than the edges. Roll one d6 and a 1 is as likely as a 6. Roll two d6 and a 7 is six times more likely than a 2, because only one combination (1+1) makes 2, while six combinations (1+6, 2+5, 3+4, 4+3, 5+2, 6+1) make 7. The more dice you pool, the more the total clusters around its average and the rarer the extreme rolls become — the same shape that shows up whenever independent random amounts are added together.

Worked examples

Sum probabilities for 2d6

By contrast, a d20 has no "middle bump" at all — every face from 1 to 20 sits at a flat 5%, which is exactly why d20 systems use a single die for pass/fail checks (any target number is easy to reason about) and reserve multi-die pools like 2d6 or 8d6 for damage and other quantities that should cluster near an average.

Common mistakes

• Applying the modifier per die instead of once. 2d6+3 is (die + die) + 3, not (die+3) + (die+3).
• Expecting a multi-die total to behave like a single die. After a string of low 2d6 rolls, a very high roll is not "due" — each roll is independent of the last (the gambler's fallacy).
• Typos in notation such as a stray space, a missing "d", or a percent sign — the field validates the pattern and flags anything it can't parse instead of guessing.
• Assuming a screen RNG needs auditing like a physical die. This roller draws from the browser's crypto.getRandomValues and discards out-of-range draws (rejection sampling) rather than using modulo, which is what a plain Math.random() % sides approach would do and which subtly favors low faces on many side counts.

What this tool doesn't do

It rolls and sums independent dice — it does not implement house rules that change the math, such as dropping the lowest die, exploding dice that reroll on a max face, or rolling twice and keeping the higher result (advantage). Any of those can still be done manually: roll the pool once for a normal result, or roll it twice and compare the totals yourself for an advantage-style check. And while a cryptographic RNG is more than fair enough for games, quizzes, and giveaways, it is not a substitute for a hardware random source in contexts that need certified, audited randomness, such as lotteries or gambling regulation.