10 Sided Dice Roller

What a 10 sided dice roller does

A ten sided die, the d10, is the workhorse of percentile systems and dice pools. The roller on this page throws one to twenty d10s at once, adds a modifier of your choice, and prints every die on screen with the total, the high and low rolls and a running session tally. Press the button or change any input and a fresh throw appears: changing the count rolls the new pool right away, while changing the modifier keeps the dice you have and updates the total.

Each die is a random number from 1 to 10. Roll three and add a +2 modifier, and the roller might show 7, 2 and 10: the dice sum to 19, and 19 + 2 = 21 lands as the final result. The same throw had an expected value of 16.5 before the modifier, because three dice at an average of 5.5 each give 3 x 5.5 = 16.5, and the spread runs from a low 3 (three 1s) to a full 30 (three 10s).

The tally line keeps the session log: every throw adds its dice to the running count, so after twenty throws of two dice the log shows 40 individual rolls with their average, and the count of 10s tells you whether the dice have been kind to you.

Reading a d10: 0 to 9 or 1 to 10

Most physical d10s print the digits 0 through 9 rather than 1 through 10. When the die is rolled alone, table convention reads the 0 as a 10, so every face still means a value from 1 to 10. When the die is rolled as part of a percentile pair, the same 0 means a zero, because 0 is a real digit of the two digit number. The roller handles both readings with a face style switch: the 1 to 10 style shows the value straight, and the 0 to 9 style shows the printed digit while the total still reads 0 as 10.

The mapping matters most in percentile play, where a 10 in the 1 to 10 style stands for the digit 0. A pair showing 10 and 7 reads as 07, just 7, not 107. The short version: the tens die 10 means zero tens, and the only way to reach the top of the scale is 0 and 0, which reads as 100.

The odds behind one d10 and a dice pool

A fair d10 gives every face the same 1 in 10 chance, so a specific number shows up 10 percent of the time, and the top face is no exception. Strings of dice obey the complement rule: the chance of at least one 10 in a pool equals 1 minus the chance that every die misses.

With two dice the miss chance is 0.9 x 0.9 = 0.81, so the hit chance is 1 - 0.81 = 0.19, or 19 percent. Three dice give 1 - 0.729 = 0.271, about 27 percent. The gains taper because each added die multiplies the miss chance by another 0.9:

Sums behave differently from individual faces. Two d10s can land on any total from 2 to 20, and the middle totals dominate: a sum of 11 comes up 10 times in the 100 ordered pairs, while a sum of 2 or a sum of 20 comes up once each. A single die has no favorite number; a pool always grows a hump in the middle. The MathWorld dice page works through the same counting for other die sizes.

Worked example: 4d10 plus 2

Many systems roll a damage hand like 4d10+2. Suppose the roller shows 4, 10, 7 and 3. The dice sum to 24, the modifier adds 2, and the final damage is 26. Before the throw you could have written the same numbers down in advance: four dice at 5.5 average give 4 x 5.5 = 22, the modifier adds 2, so the average result is 24, the floor is 4 + 2 = 6 and the ceiling is 40 + 2 = 42.

A second example shows how the spread grows with pool size. A single d10 plus 5 lands between 6 and 15, a range of 10 values, with the average at 10.5. Four d10s plus 5 lands between 9 and 45, a range of 37 values, with the average at 27. The average grows in clean steps of 5.5 per extra die, while the spread grows roughly with the square root of the dice count, so bigger pools feel steadier even though their extremes are wider.

When a system wants a bell shaped result from d10s, the pool itself is the tool: 3d10 replaces 1d30 with a distribution peaked at 16.5 instead of a flat line, and the extremes 3 and 30 each come up about once in a thousand throws.

Percentile rolls with a pair of d10s

Two d10s make a d100. One die is the tens digit, printed 00 to 90 on real percentile dice, and the other is the ones digit. The roller reads the first two dice of any throw as a percentile pair: a die showing 7 in the tens place and 4 in the ones place gives 74, and every number from 1 to 100 shows up exactly once per hundred ordered pairs, a flat 1 percent per value.

The pair 00 and 0 is the famous one. Read as digits it would be zero zero, but table convention gives the player the top of the scale, 100, on the same logic that makes a single 0 into a 10. Without the convention the pair would never produce 100 and the distribution would miss its last value.

Percentile rolls power random tables, treasure generation and any effect described as a percentage. With the 1 to 10 face style the roller maps a 10 to the digit 0 automatically, so 70 and 10 still reads as 70, and 10 and 10 reads as 100. Check the reading in the results table before you apply a table result; the percentile line prints on every throw of two or more dice.

Dice pools that count 7s and up

Dice pool systems roll a handful of d10s and count successes instead of adding the faces. The classic target is 7 or higher, which on a d10 covers the faces 7, 8, 9 and 10, so each die succeeds 4 times in 10, a 40 percent success rate. Expected successes are pool size times 0.4: a pool of 3 expects 1.2, a pool of 5 expects 5 x 0.4 = 2.0, and a pool of 8 expects 3.2. The roller supports the style directly: roll the pool, read the dice line and count the faces at 7 or above.

Botch rules lean on the same die. If a 1 cancels a success and a throw with no successes and at least one 1 is a disaster, the odds sit on the low face: a single die botches 10 percent of the time, and the all 1s outcome across five dice lands once in 100,000 throws. Some systems let 10s count double, which nudges the average up without changing the 40 percent base rate, and the roller's tens counter tracks how often the top face appeared in the session.

Fair numbers and rolling without a d10

The roller draws from crypto.getRandomValues, the browser's cryptographic random source, and discards any draw that would bias the last digit, so all ten faces stay equally likely. When that source is unavailable the code falls back to Math.random, which is uniform enough for table play. The NIST randomness guidance (SP 800-90B, 2018) is the reference behind this style of rejection sampling. Nothing leaves the browser: the dice, the log and the totals never touch a server.

Without any device at hand, several stand ins produce a fair 1 to 10. A d20 halves cleanly: roll it and treat 1 or 2 as 1, 3 or 4 as 2, up to 19 or 20 as 10, so every value gets two of the twenty faces. A deck of cards with the face cards removed gives A through 10. A spinner divided into ten equal sectors works, and a pair of differently colored d10s keeps working as the percentile pair described above. Physical d10s wear unevenly, and a die that has been rolled thousands of times can drift away from flat; the digital roller cannot, which is one reason online tables moved their dice to the browser.