Behind the birthday odds

The U.S. birth records behind the results, what each number means, and how we account for the calendar.

85,712,738 recorded births from 1994–2014.

The checker, date rankings, and month comparisons use historical U.S. birth counts from 1994 through 2014, compiled by FiveThirtyEight from CDC/NCHS and Social Security Administration records. We combine them into one set of 366 month/day totals for the checker, rankings, and charts.

The selected records cover every calendar day from January 1, 1994, through December 31, 2014: 7,670 daily observations, with no missing dates. See the source files and coverage.

Your birthday rank

The main checker ranks dates by their total number of births. Rank #1 means the most births; rank #366 means the fewest when all dates have different counts. A higher rank number means a less common birthday.

Rank = 1 + the number of dates with strictly more births

Dates with equal counts share a rank. If two dates tie for first, the next date ranks third. On the ranking pages, switching to births per occurrence recalculates ranks using that measure instead.

Share of births and “1 in…”

A date’s share is its birth count divided by all births in the dataset. If 20 out of 10,000 recorded births fell on a date, its share would be 0.2%, or 1 in 500 recorded births. This is a mathematical example, not a measured finding.

Share = births on the date ÷ all births
“1 in X” = all births ÷ births on the date

Displayed numbers are rounded for readability. The calculations use unrounded values. Historical birth shares describe the records being counted; they are not a forecast of a particular baby’s birth date.

What “rarer than” means

This percentage compares birthday dates. “Rarer than 80% of other birthday dates” means 80% of the other dates have more births. It does not mean an 80% probability, and it does not compare you with 80% of people.

Rarer than % = dates with strictly more births ÷ other observed dates × 100

Equal counts do not count as strictly more common. The selected date is excluded from its own comparison.

February 29 needs context

February 29 occurred 5 times in 1994–2014; every other month/day occurred 21 times. A ranking by total births includes that calendar effect. A non-leap year does not contribute a zero-birth February 29; the date does not occur in that year.

Births per occurrence divides a date’s total by the number of times it actually appeared. This answers a different question: how many births happened on an average occurrence of that date?

Explore both measures on the rare birthdays page.

Putting months on equal footing

Monthly totals show how many births fell in each month. To compare daily frequency, we divide each total by the number of actual covered days in that month, including leap days. That keeps a longer month from getting an automatic advantage.

Compare monthly totals and daily averages.

Holiday and weekday research

The holiday comparison and weekday analysis use SSA records from 2000 through 2014 alone. Using one series avoids mixing administrative sources within those comparisons. Their coverage differs from the checker and main guides.

Weekday averages divide births by the number of actual days in each group. The holiday study compares four selected fixed dates with nearby dates on the same weekday, reports exclusions at the edges of the coverage period, and checks how results change when comparison dates are themselves holidays. Each article gives its full method and limitations.

Download the research and guide comparison data. The file includes source coverage, counts, denominators, and alternative holiday comparisons.

Shared-birthday probability

The birthday paradox calculator compares two models for the chance that at least two people in a group share a month and day. Both assume each person's birthday is independent of the others.

The classic model gives each of 365 dates the same chance and excludes February 29. The historical model uses each of the 366 dates' shares of 85,712,738 recorded U.S. births from 1994 through 2014, including February 29. For both, we calculate the chance that all birthdays differ and subtract it from 1. The historical weights use total birth shares, rather than births per calendar occurrence.

These are modeled probabilities, not observed matching rates in actual social groups. The comparison changes date weights and leap-day treatment together. Display rounding can make a very high probability look certain; the calculator shows 100% only when there are more people than possible dates in the chosen model.

Download the probability calculations and September ranking comparison. The article explains the separate question of someone matching your own birthday and links to the mathematical derivation.

Sources and attribution

The original daily files are published in FiveThirtyEight’s U.S. births dataset with a description of each source. CDC/NCHS refers to the Centers for Disease Control and Prevention’s National Center for Health Statistics; SSA is the Social Security Administration.

The files overlap in 2000–2003. We use only SSA for those years, excluding the overlapping CDC/NCHS rows so no calendar day is counted twice. We preserve the original files and calculate the month/day and monthly summaries used on this site.

The files were retrieved on from revision 4c1ff5e3aef1 of the upstream repository. The retrieval date records when we obtained the files; the birth records end in 2014.

FiveThirtyEight makes its datasets available under Creative Commons Attribution 4.0 International (CC BY 4.0) unless otherwise noted. Its births documentation lists no separate exception. Attribution: U.S. birth counts from CDC/NCHS and SSA, compiled by FiveThirtyEight. Birthday Odds selects CDC/NCHS 1994–1999 and SSA 2000–2014 and calculates month/day and monthly summaries.

What these records can tell us

These results describe recorded U.S. births during 1994–2014. They do not describe worldwide birthdays, current birth rates, or everyone alive today. A visitor’s date is looked up in these historical records; visitor selections never change the rankings.

Administrative coverage differs between CDC/NCHS and SSA. Combining their counts gives us a longer period to explore, but a change between the two sources cannot be assumed to be a population trend. Weekday patterns and the years selected can also affect comparisons.

A pattern on the calendar does not establish its cause or explain why an individual was born on a particular date.