Dice Notation
The difference between a d20 and a d100 table
One die with a hundred faces and two ten-sided dice read as digits produce the same range and different habits. A look at why systems pick one, and what changes when they do.
A range of a hundred is often produced a different way from a single hundred-faced die: you roll two ten-sided dice and read one as the tens digit and the other as the units digit. Many rulebooks that need a range of a hundred use that reading.
The two methods give the same nominal range. They do not give the same object at the table, and the difference is worth understanding before you trust a percentile table. This note also builds one small table of our own: which percentile tables a d20 can represent exactly, and which it cannot.
How two ten-sided dice make a hundred
A ten-sided die is written 1d10 in the free rules’ notation, a count followed by d and the number of sides (D&D Beyond — Free Rules: Playing the Game). Two of them have 10 × 10 = 100 ordered outcomes, each with probability 1/100 when the dice are fair. Listing the outcomes of a pair of fair dice as ordered pairs is the same move OpenStax — Mean or Expected Value and Standard Deviation makes with two tosses of a six-sided die, where the 36 pairs are written out. A fair hundred-faced die would have the same distribution, one face per value, so on paper the two are interchangeable.
The free rules spell out the reading rule. The common way to roll a d100 is a pair of ten-sided dice numbered 0 to 9; you decide before rolling which die gives the tens digit and which gives the ones, so a 7 and a 1 read as 71, and two 0s count as 100. Some ten-sided dice are printed in tens (00, 10, 20 and so on), which makes the two easier to tell apart. Other games may read 00 differently, so check the convention in the book you are using.
Where the two methods differ in the hand
These are practical differences, not probability differences.
- Which die is which. With two ten-sided dice a tired player can swap the tens and the units, and the result jumps by tens. The free rules ask you to name the tens die before rolling; a pair of different colours helps too.
- Boundary reading. The free rules read 00 as 100, so a table runs from 1 to 100; a game that reads 00 as 0 shifts the top and bottom of every table by one value. State the convention once, next to the table.
- The feel of the roll. Throwing two dice and reading a number feels like arithmetic. A game that wants players to feel they are working with a percentage will usually choose the pair.
Reading a percentile table
Percentile tables are usually built as a ladder: roll, read the number, look up the band. A table reading “01–05: catastrophic”, “06–40: failure”, “41–85: ordinary success”, “86–96: strong success”, “97–00: exceptional” is a set of five intervals over a hundred equally likely values. Three checks make such a table readable at speed:
- Total the bands. These five bands cover 5 + 35 + 45 + 11 + 4 = 100. A table that does not total exactly 100 has a gap or an overlap, and it will be the row nobody can find at speed.
- Read the boundaries as inclusive. “41–85” includes 41 and 85. A table where one band ends at 40, the next starts at 41, and another starts at 86 right after a band that already used 86 has two overlapping rows.
- Know what 00 means. If the book reads 00 as 100, say so in the same box as the table.
What a d20 can and cannot stand in for
Each face of a d20 covers five percentile values (5% each). A percentile table can therefore be run on a d20 only if every band size is a multiple of 5. Here is the example table above, checked against that rule:
| Band | Values | Size | Size ÷ 5 | Exact on a d20? |
|---|---|---|---|---|
| Catastrophic | 01–05 | 5 | 1 | Yes: 1 face |
| Failure | 06–40 | 35 | 7 | Yes: 7 faces |
| Ordinary success | 41–85 | 45 | 9 | Yes: 9 faces |
| Strong success | 86–96 | 11 | 2.2 | No |
| Exceptional | 97–00 | 4 | 0.8 | No |
The first three bands convert exactly (a roll of 1, then 2 to 8, then 9 to 17). The last two cannot: 11 values would need 2.2 faces and 4 values would need 0.8 of a face. Converting would force the designer to move the boundaries, for example to bands of 10 and 5, which is a different table. That is the real cost of swapping a d100 table for a d20 table: you keep the range of the dice but lose the fine grain of the bands.
The d20 has its own strength. A d20 cannot express a 63% chance without a target number, but a target number is easy to adjust at the table. A percentile roll answers “how likely is this?” in one step. A d20 answers “does this beat a threshold?” in one step. The choice of die follows from which question your game asks most.
Method
The band sizes and the 5-value-per-face rule come from counting: 100 percentile values divided by 20 faces. The band sizes were summed by script (5 + 35 + 45 + 11 + 4 = 100) and each was divided by 5. The band labels and ranges are an illustrative table we wrote for this note, not taken from a published game. The dice notation was read on 2026-10-31 on the page listed below; the same page describes percentile dice (a pair of ten-sided dice, tens and ones, two 0s as 100), which the reading rule above follows; conventions of other games are described as things to check in your own book, not as rules.
Sources
- D&D Beyond — Free Rules: Playing the Game: dice written as a count, d and the number of sides, and the percentile-dice reading rule.
- OpenStax — Mean or Expected Value and Standard Deviation: listing the ordered outcomes of two dice as a sample space.
Written for: a reader who has seen percentile tables in a rulebook and wants to know whether the choice of die changes anything that matters.
Sources
- D&D Beyond — Free Rules: Playing the Game (d20 Tests and dice notation) — accessed 2026-10-31
- OpenStax — Mean or Expected Value and Standard Deviation — accessed 2026-10-31