Age Calculator

Why calendar arithmetic is harder than it looks

Working out an age sounds like subtraction and is not. Subtracting the years gets you close and is wrong for roughly half the year, and the correction is where every disagreement between two calculators comes from.

Months are not the same length

The obvious way to do this is to subtract the day numbers and borrow when the answer goes negative, the way you would with columns of figures. It works most of the time and then breaks in a way that is easy not to notice.

Take somebody born on 31 January, measured on 1 March. Subtracting gives 1 − 31 = −30, so you borrow a month. But February only has 28 days to lend, which leaves you at −2 — still negative, and the answer is nonsense. Borrowing a flat 30 days, as a lot of code does, gives "1 month and 0 days", which is wrong in a quieter way.

The rule that holds up is different: add whole months to the start date until one more would overshoot, then count the days from there. 31 January plus one month is 28 February — a month has passed, even though there is no 31st of February — and from there to 1 March is one day. One month and one day.

That is what this calculator does, which is why the same number of days can come out as a different number of months depending on which months it crosses.

What "one month later" means

There is no universal answer. One month after 31 January is 28 February in most systems, because 31 February does not exist, and the convention is to clamp to the end of the month. But then one month after 31 January and one month after 28 January are the same date, which means the operation cannot be reversed.

This is not a flaw in any particular calculator. It is a property of a calendar with months of unequal length, and every piece of software has to pick a convention.

Being born on 29 February

About one person in 1,461 is, and three years in four their birthday does not exist.

The common convention, used here, is that the anniversary falls on 1 March in non-leap years — you have completed a full year at the moment the equivalent date would have passed. Several legal systems agree; a few, including some in Asia, use 28 February instead so that the birthday stays inside the birth month.

For an age in whole years the difference is one day, which matters exactly once in most lives and then never again.

Why leap years exist at all

A year is about 365.2422 days, not 365. Adding a day every four years overcorrects slightly, so century years are not leap years — except every fourth century, which is. 1900 was not a leap year; 2000 was. That last rule catches a great deal of software, which is why some old systems think 29 February 2000 did not happen.

Ages are not counted the same way everywhere

The age described here is the international one: you turn a number on your birthday and are that number until the next.

Traditional East Asian age reckoning counts differently — a baby is one at birth and gains a year at the new year rather than on a birthday, which makes people one or two years older than the international count. South Korea moved officially to the international system in June 2023, but the traditional count is still used conversationally and in some contexts.

Time zones, and why this uses none

Your birthday is a date, not a moment. Somebody born in Tokyo who moves to Buenos Aires does not get older twelve hours early.

So this calculator works in whole local dates and never converts to UTC. That sounds obvious and is a common bug: parsing "1990-06-15" as a timestamp gives midnight UTC, which is the evening of 14 June for anybody in the Americas — and produces an age one day out for half the planet.