How Roman numerals work

The seven symbols

Roman numerals use seven letters, each with a fixed value. Everything else is built by combining these.

Rule one: add from large to small

Write the biggest values first and add as you move right. VIII is 5 + 1 + 1 + 1 = 8. LX is 50 + 10 = 60. MCC is 1000 + 100 + 100 = 1200. As long as each symbol is the same size or smaller than the one before it, you simply sum them.

Rule two: subtract when small comes first

To avoid four of the same letter in a row, a smaller symbol placed before a larger one is subtracted. IV is 5 minus 1, which is 4. IX is 10 minus 1, which is 9. XL is 50 minus 10, which is 40. Only I, X and C subtract, and only before the next one or two larger symbols. You will never see IL or IC; those are written XLIX and XCIX.

The six subtractive pairs in full

There are exactly six. Not five, not seven, and no others are legal in the standard system. Each one exists to keep a letter from appearing four times in a row, and each one is read as a single unit rather than as two separate letters.

The pattern behind the six is simple once you see it: only I, X and C ever subtract, and each may sit in front of just the next two larger symbols. I goes before V and X, X goes before L and C, C goes before D and M. That is three letters times two positions, which is where the number six comes from. V, L and D never subtract at all, which is why IL, VC and LM are not numerals, however sensible they look.

Reading a longer numeral

Split it at each place value. Take MCMXCIV: M is 1000, CM is 900, XC is 90, IV is 4. Add them and you get 1994. Working left to right in groups is exactly how the converter on this site parses your input.

Common mistakes

• Writing IIII for four instead of IV, except on clock faces where IIII is traditional.
• Repeating V, L or D. These never double up, so 10 is X, not VV.
• Over-subtracting, such as writing IC for 99. The correct form is XCIX.
• Forgetting that there is no zero and no fraction in the standard system.

Why there is no zero

Roman numerals have no symbol for nothing, and that is not an oversight. The system is additive: a numeral is a running tally of quantities you can see, so a quantity of none simply has nothing to write down. A Roman accountant recording an empty column wrote the word nulla, Latin for none, or left the space blank. The medieval computists who used numerals for calendar tables sometimes abbreviated that word to N, which is the closest thing to a Roman zero that ever existed, and it was a word standing in for a number rather than a numeral in its own right.

The practical consequence is that Roman numerals cannot express place value. In 1994 the Arabic 0 does real work: it says that the tens column is empty while still holding the hundreds column in place. MCMXCIV has no such mechanism, which is why every group has to be written out in full and why arithmetic in numerals is so laborious. It is also why the system stops being useful long before it stops being writable.

Above 3999: the vinculum and the bracket form

MMMCMXCIX is 3999 and it is the largest number the seven letters can express on their own. You cannot write 4000 as MMMM, because no symbol repeats more than three times.

The Roman answer was the vinculum: a bar drawn over a numeral to multiply it by a thousand. A V with a bar over it is 5000, an X with a bar is 10,000, and a numeral with two bars is multiplied by a million. Because a bar cannot be typed reliably in plain text, the common modern substitute is a bracket, so 5000 is written (V) and 10,000 is written (X). That is the form the converter on this site produces above 3999, and it is a typographic convention rather than something Rome itself standardised. Where a claim like that is a convention rather than a hard fact, we say so.

Where the system came from

The origins are less settled than the arithmetic, so this section is deliberately cautious. The usual account is that the numerals began as tally marks rather than as letters: a notch for one, a deeper or angled notch for five, a crossed notch for ten, which over time were assimilated to the Latin letters they resembled. That would explain why I, V and X behave like counting marks while L, C, D and M behave like abbreviations.

C for 100 and M for 1000 do match centum and mille, and that correspondence certainly helped the symbols stick, though most reference works treat it as a later rationalisation rather than the origin. L for 50 fits no Latin word at all and is generally traced to an Etruscan sign, which is the clearest hint that the set was inherited rather than designed.

The subtractive notation is later than the letters themselves. Roman inscriptions regularly show IIII and XXXX, and the strict six-pair system taught today was only fixed in the medieval and early modern period. That is the honest reason clock faces still show IIII: they preserve a form that was never wrong, only superseded.

How to check a numeral is canonical

Knowing the rules is one thing, catching your own mistake is another. There is a check that takes ten seconds and does not require you to trust anyone, including this site. Convert the numeral to a number, then convert that number back into a numeral. If the result is not identical letter for letter to what you started with, the original was not canonical.

IIII reads as 4, and 4 written properly is IV, so the round trip fails and you have your answer. XCIX reads as 99 and 99 written properly is XCIX, so it survives. VIIII reads as 9 and 9 is IX, so it fails. IL is not a valid numeral at all, because I may only precede V and X, so there is nothing to read back. This is exactly what happens inside the converter on this site before any result is shown, which is why a rejected form never reaches the page.

It matters most in the situations this site was built for. A numeral going onto a ring, a headstone or someone's skin is not a draft. Run the round trip once before it becomes permanent, and check the number too, not only the letters: the commonest error in a date tattoo is not a malformed numeral but a correctly formed numeral for the wrong year.

Sources

The arithmetic on this page is generated and checked by the same function that powers the converter, so it cannot drift. The historical notes are cross-checked against standard reference works rather than invented:

• Encyclopaedia Britannica, Roman numeral, for the symbol set, the subtractive rule and the tally-mark account of the origins.
• How we check accuracy, for the method: every conversion round-tripped, malformed forms rejected, cautious wording wherever the history itself is uncertain.

Frequently asked questions

Common questions

A few specific questions come up again and again once the rules click. Each has its own worked answer:

Put it into practice

Last updated July 22, 2026. Reviewed for accuracy against the standard rules for Roman numerals. How we check accuracy.