What Does Dice Notation Like 2d6+3 Actually Mean?

Dice notation NdM+K means roll N dice, each with M sides, sum their results, then add the flat modifier K. 2d6+3 means roll two six-sided dice, add the two results (a value from 2 to 12), then add 3 — giving a total between 5 and 15, most often near 10.

A worked example with numbers

Rolling 2d6 has 36 equally likely combinations, one for every pair of die faces:

How to set the rule before the result

Read the notation left to right: the number before “d” is how many dice to roll, the number after is how many sides each die has, and anything after a “+” or “−” is added once at the end, not per die. “4d6” means roll four separate six-sided dice and add all four results (range 4–24); it is not the same instruction as rolling one d6 four separate times unless those four values are then combined. A common variant, “roll 2, keep the higher,” describes advantage rather than plain notation — the advantage/disadvantage roller implements that keep-the-best rule separately from a straightforward sum. When implementing a dice roller that must support several of these notations, it helps to separate the parser (turning “2d6+3” into the numbers 2, 6, and 3) from the resolver (deciding whether those numbers mean sum, keep-highest, or count-successes), since conflating the two stages is the most common source of a roller that parses a formula correctly but resolves it the wrong way.

Common mistakes that change the odds or the process

The most common error is treating 2d6 as equivalent to 1d12, because both nominally range from 2 (or 1) to 12. They are not the same distribution: a single d12 gives every value from 1 to 12 an equal 1/12 (8.33%) chance, while 2d6 concentrates around 7 and gives the extremes of 2 and 12 only a 2.78% chance each. A second error is applying the modifier per die instead of once to the total — “2d6+3” is not the same as adding 3 to each individual die. A third is forgetting that “drop lowest” notation, such as “4d6 drop lowest,” removes one rolled die from the sum before any modifier is applied. A fourth, easy-to-miss error is misreading which die was dropped when two dice tie for lowest: with a roll of 3, 5, 3, and 6 under “4d6 drop lowest,” either of the two 3s can be the one removed without changing the kept sum of 5 + 3 + 6 = 14, but a script that drops the wrong die entirely — say, the first 5 instead of a 3 — silently changes the result and needs its own test case built around a tie.

Where this method stops being appropriate

Standard NdM+K notation assumes independent, fair dice and a simple additive rule; it does not by itself define exploding dice, reroll-on-1 rules, critical-hit doubling, or a dice pool that counts successes instead of summing values — those are separate house rules layered on top of the base notation, and a plain dice roller will not apply them unless it explicitly implements that mode, such as the dice pool roller. It also assumes every die in the notation is fair and correctly sized; a physical die that has been damaged, a mislabelled custom die, or a digital implementation that generates the wrong range for “d20” will make the notation’s stated probabilities describe an ideal case rather than the actual results a player sees.

How the random source fits into the rule

the D&D Beyond Basic Rules documents standard die notation and how modifiers combine with rolled dice in tabletop rules, which is the convention the NdM+K shorthand is built on. The same source also documents advantage and disadvantage as an explicit alternative to plain notation, which is why a rules-accurate implementation treats “roll with advantage” as a separate instruction rather than a variant spelling of a normal die roll.

Why 2d6 is not the same as rolling one d12

Both methods can produce any integer from their respective minimum to their maximum, which is why they are sometimes assumed to be interchangeable. The table shows they are not: a d12 spreads probability flat across all twelve faces, while 2d6 piles probability toward the middle because more face combinations add up to 7 than to 2 or 12. A game rule that specifies “roll 2d6” is deliberately choosing the bell-shaped version.

The same distinction scales up: 3d6 (used for classic tabletop ability scores) ranges from 3 to 18, with 10 and 11 each the most likely totals at roughly 12.5%, while a single d16 relabelled 3–18 would spread that same range flat at 6.25% per value. Notation that looks superficially similar on paper — “roughly the same range” — can describe very different games once the actual distribution shape is compared.

How advantage and dice-pool notation build on the same rule

Advantage — rolling two dice and keeping the higher — is not additive notation at all; it changes which single die counts, which is why its probabilities differ from both 1d20 and 2d20 summed, as worked out in advantage and disadvantage in dice games. A dice pool such as rolling six ten-sided dice and counting each 8 or higher as a success replaces the sum entirely with a count, using the same independent-dice assumption but a different final calculation. Reading any notation correctly means identifying which of these three operations — sum, keep-highest, or count-successes — the dice are actually being used for, since a shorthand like “6d10” alone does not say which one applies without the surrounding rule text.

A useful habit when reading an unfamiliar ruleset is to find the one sentence that defines what happens to the individual die results before treating any shorthand as self-explanatory: “sum all dice,” “keep the highest,” “count values at or above a threshold,” and “drop the lowest N” are four different operations that can all be written with similar-looking Nd M notation, and only the accompanying rule text distinguishes them.

The full 2d6 probability table

All eleven rows total exactly 36 ways and 100% probability, a useful arithmetic check on any dice table before trusting it: 1+2+3+4+5+6+5+4+3+2+1 = 36. Applying the +3 modifier from “2d6+3” shifts every row’s label up by three — 2 becomes 5, 7 becomes 10, 12 becomes 15 — while every probability in the right-hand column stays exactly where it is.

A second case: reading 3d6+2 the same way

The same left-to-right reading rule applies to any NdM+K string with more dice. “3d6+2” means roll three six-sided dice, sum them — a value from 3 to 18 — then add 2 once, giving a total from 5 to 20.

Three summed dice spread across 216 equally likely combinations rather than 36, and the distribution concentrates even more tightly around its middle than 2d6 does: the two most common raw totals, 10 and 11, share exactly 12.50% each, while the two extremes, 3 and 18, each occur only 1 time in 216 — about 0.46%, noticeably rarer than 2d6’s already-rare 2.78% extremes.

Why “4d6 drop lowest” is not “4d6 minus one die”

A frequent mistake treats “4d6 drop lowest” as subtracting one average die from a plain 4d6 sum: 4 × 3.5 − 3.5 = 10.5. That shortcut is wrong because the dropped die is not a randomly chosen one — it is specifically whichever of the four came up lowest, so the removal strips away more low results than an average die would. Enumerating all 1,296 possible outcomes gives a true expected value of approximately 12.24, nearly two full points higher. “Drop” or “keep” notation describes a selection rule applied before summation, not a simple arithmetic adjustment to a plain sum, and the two calculations can disagree widely.

How many rolls it takes to confirm a rare row yourself

The rarest row in the 2d6 table, a raw sum of 2 at 2.78%, is also the easiest one to misjudge from a small sample: a handful of rolls without a 2 proves nothing, and a single 2 in ten rolls does not mean the true rate is 10%. Pinning that 2.78% figure down to within ±0.5 percentage points at 95% confidence takes roughly n = 1.96² × (1/36)(35/36) ÷ 0.005² ≈ 4,150 rolls of 2d6, each recorded individually — well beyond what a single game session produces, which is why a suspiciously rare-feeling roll is better checked against a documented probability table than against a short private tally.

What the notation cannot verify on its own

Standard NdM+K notation defines what a correct result should look like statistically; it says nothing about whether a specific dice app or virtual tabletop actually implements that rule correctly. A “4d6 drop lowest” button that silently drops the wrong die on a tie, or a “2d6+3” field that adds the modifier per die instead of once, still looks like a plausible dice result at a casual glance. Confirming an implementation matches its notation needs the same large-sample check as confirming a physical die is fair — thousands of logged results compared against the table, not a handful of plausible-looking rolls.

Related DiceDecide tools

• Roll a two-dice sum directly
• Try advantage and disadvantage rolls
• Count successes across a dice pool
• Read the full probability behind advantage