Why 23 People Are Enough for a Shared Birthday to Be Likely
Before you read any further, have a guess.
Imagine a room full of people. How many would you need before it's more likely than not that two of them share a birthday? Same day, same month, year doesn't matter.
There are 365 days to choose from. So it feels like you'd need a lot of people. A hundred? Maybe more? Half of 365 is about 180, so that sounds like a reasonable answer.
The real answer is 23.
Just over half
With 23 people in a room, the chance that at least two of them share a birthday is about 50.7 per cent. Just over a coin flip.
And it climbs quickly from there. By the time you have 70 people, the chance of a shared birthday is about 99.9 per cent. Very nearly certain.
Think about that next time you're at a school assembly, or a wedding, or watching two football teams line up with the referees. Almost any gathering of that size is likely to contain a birthday match.
So why does 23 feel so wrong?
People versus pairs
The trick is in what you're counting.
When most of us think about this question, we picture ourselves in the room. We imagine checking our own birthday against everyone else's. In a room of 23, that's 22 other people, and 22 chances for a match out of 365 days doesn't sound like much.
But the question isn't whether someone shares your birthday. It's whether any two people share a birthday. And that's a very different count.
Every single person in the room can be compared with every other. The first person can be paired with 22 others. The second person with 21 more that haven't been counted yet. Then 20, then 19, and so on, all the way down to 1.
Add those up, and 23 people make 253 different pairs. Each one of those pairs is a little lottery ticket for a shared birthday.
That's the heart of the whole thing. People go up one at a time. Pairs go up much faster.
Working it out backwards
Counting all the different ways a match could happen gets messy. So mathematicians do something sneaky. They work out the chance that there's no match at all, and take that away from 100 per cent.
Here's how it goes, in words.
Start with the first person. Whatever their birthday is, it's fine. No one to clash with yet.
Now the second person walks in. For there to be no match, they have to have a birthday different from the first person's. That leaves them 364 safe days out of 365. Very likely.
The third person has to avoid two birthdays that are already taken. That leaves 363 safe days.
The fourth person has to avoid three. And so on.
Each new person has a slightly smaller set of safe days, and for there to be no match at all, every single one of them has to land on a safe day. Each step on its own is very likely to go fine. But you're multiplying all those chances together, and those small odds of a miss stack up surprisingly fast.
By the time the 23rd person walks in, the chance that everyone has dodged everyone else has fallen to just under half. So the chance of at least one match is just over half.
Your birthday is still special
Here's where it gets genuinely confusing, so let's slow down.
None of this means that someone in a room of 23 is likely to share your birthday. They almost certainly won't. The chance that at least one of the other 22 has your exact birthday is only about 6 per cent.
In fact, if you wanted a better than even chance of finding someone with your own birthday, you'd need around 253 other people.
So both of these things are true at once. In a room of 23, your own birthday is very probably unique. And yet a shared birthday somewhere in the room is more likely than not.
That's exactly why it fools us. Our instinct answers the question about ourselves. The maths answers the question about everyone.
Coincidences in general
This turns out to be a much bigger idea than birthdays.
In 1989, the statisticians Diaconis and Mosteller wrote about the mathematics of coincidences in the Journal of the American Statistical Association. One of the principles they set out has become known as the law of truly large numbers. The idea is that, with a large enough number of people and opportunities, events that seem astonishing are not just possible but expected. In a line quoted by Wolfram MathWorld, they wrote that "we can be absolutely sure that we will see incredibly remarkable events".
The birthday problem is a small, neat example of that. A coincidence that looks spooky is often just the result of many more chances than we realised.
The same maths turns up in computer security, too. Systems that turn data into short codes can occasionally produce two identical codes from different inputs, a clash called a collision. Because of the birthday effect, those clashes turn up far sooner than you might naively expect, and attacks that take advantage of this are known, fittingly, as birthday attacks.
The small print
The classic figure of 23 rests on some simplifying assumptions.
It assumes there are 365 possible birthdays, each equally likely, and that everyone's birthday is independent of everyone else's. It ignores leap days. It ignores twins. It ignores the fact that more babies are born at some times of year than others.
Real birthdays aren't perfectly evenly spread. But here's the nice bit: unevenness actually makes a match slightly more likely, not less. If some days are more popular, people are more likely to bump into each other on them. Real-world birthday patterns aren't uneven enough to change the magic number, though. It's still 23.
Counting the wrong thing
So why are 23 people enough?
Because coincidences don't come from people. They come from pairs of people, and pairs multiply much faster than our intuition expects.
Coincidences aren't as rare as they feel. Our sense of "rare" is just counting the wrong thing.
Sources
Science Buddies, Scientific American, "Using Probability to Understand the Birthday Paradox" (Bring Science Home, 29 March 2012). https://www.scientificamerican.com/article/bring-science-home-probability-birthday-paradox/ (accessed 1 Nov 2026)
P. Diaconis and F. Mosteller, "Methods for Studying Coincidences", Journal of the American Statistical Association 84 (December 1989), pp. 853–861. https://www.jstor.org/stable/2290159 (quote via MathWorld)
Wolfram MathWorld, "Law of Truly Large Numbers". https://mathworld.wolfram.com/LawofTrulyLargeNumbers.html (accessed 1 Nov 2026)
Wikipedia, "Birthday problem". https://en.wikipedia.org/wiki/Birthday_problem (accessed 1 Nov 2026)
M. Cortina Borja and J. Haigh, "The birthday problem", Significance 4(3) (2007). doi:10.1111/j.1740-9713.2007.00246.x (citation supplied by Reed, for real-world birthday clustering)
