Probability/3 min exploration

A room full of coincidences

How quickly does a shared birthday become more likely than not?

ProbabilityMAKE A PREDICTION

How many people does it take for the chance of any shared birthday to rise above 50%?

Find the smallest group. Assume 365 equally likely birthdays, chosen independently, with no leap day. A match between any two people counts.

YOUR ESTIMATE
50PEOPLE1,225POSSIBLE PAIRSAny pair can share a birthday.
Each dot is one person. Lines illustrate some pair comparisons; the total counts every possible pair.
50 people
2 people100 people

A rough guess is enough.

The bigger picture Probability model · reviewed 4 October 2026

What the numbers mean

A small room can contain a lot of possible pairs.

In a group of 23, there are 253 different pairs of people. Under this model, the chance that at least one pair shares a birthday is about 50.7%. With 22 people, it is still below half, at about 47.6%.

Read the sources
01

Count pairs, not just people

If you look only for someone matching your birthday, there are 22 possible matches in a group of 23. But this question includes everyone matching everyone else. Count each pair once: 23 × 22 ÷ 2 = 253. The number of possible connections grows much faster than the size of the group. [1]

02

It is easier to count no matches

The first person can have any birthday. The next must avoid that day, leaving 364 out of 365 possibilities. The third must avoid two days, and so on. Multiplying those chances gives the probability of no shared birthdays. Subtracting it from one gives the chance of at least one match. [2]

03

More likely still leaves room for surprise

A 50.7% chance is far from a guarantee. Almost half of groups of 23 would have no match under this model. Real birthdays also vary by season, and leap day adds a complication. The model explains the pattern; it does not predict the birthdays in a particular room. [2]

Sources & method

How we worked it out.

Probability model · reviewed 4 October 2026

Sources checked

For n people, multiply 365/365 × 364/365 × ... × (366 − n)/365, then subtract from 1. The first whole-number group size above a 50% match probability is 23. Birthdays are independent and uniform across 365 days in this model. Pair counts are n × (n − 1) ÷ 2; the pair events themselves are not all independent.

  1. 01MIT OpenCourseWareMathematics for Computer Science: the birthday paradox, section 16.5 (2010 PDF) (opens in a new tab)
  2. 02Charles Grinstead and J. Laurie Snell / Dartmouth CollegeIntroduction to Probability: the birthday problem, pages 79–80 (PDF) (opens in a new tab)