Skip to main content
  • Pricing
Tutorials

Why the number 37 keeps showing up: random picks, prime facts and the 37% rule

By Oakgen TeamPublished 3 min read

Share
Why the number 37 keeps showing up: random picks, prime facts and the 37% rule

The number 37 draws attention for three separate reasons, and it helps to keep them apart. It has tidy arithmetic properties, it is the cutoff in a well-known stopping rule when you have 100 options, and many people say it feels like a "random" pick. Only the first two follow from mathematics. The third is a claim about human behavior, and the evidence for it is thinner than the popular retellings suggest.

What is plainly true of 37

Wolfram MathWorld describes 37 as the 12th prime number, the fourth hex number and the third star number. The primality check is quick by hand. The square root of 37 is about 6.1, so only the primes 2, 3 and 5 need testing. 37 is odd, its digits sum to 10 (so 3 does not divide it), and it does not end in 0 or 5. That confirms it is prime.

Two small identities explain why 37 appears in repeating decimals:

  • 3 x 37 = 111, so 37 divides 111, and also 111,111 (which is 111 x 1001).
  • 27 x 37 = 999, so 1/37 = 27/999 = 0.027027027... repeating every three digits.

These are calculations you can verify with a pencil. MathWorld itself is explicit that the prominence of 37 in requests for a supposedly random two-digit number is "a psychological and cultural phenomenon, not a consequence of these arithmetic properties."

What the research literature says about number choices

The classic result concerns 7, not 37. A University of Michigan statistics textbook, citing a 1976 paper by Kubovy and Psotka in the Journal of Experimental Psychology: Human Perception and Performance, describes students stopping 558 passersby on the Yale campus and asking for the first number that came to mind between 0 and 9. About 28 percent chose 7. A 1972 study in Psychological Reports by Simon and Primavera asked 533 elementary and junior high school children for a number from 0 to 9 and a color name, and found 7 and blue were by far the most frequent, consistent with an earlier study of college students.

A broader point comes from lottery data. Researchers analyzing real entries from two lottery games (more than five million combinations in the first) reported that number choice is "far from uniform," that players favor personally meaningful and situationally available numbers, and that they are drawn toward numbers in the center of the choice form. Human number picking, in short, has strong, repeatable patterns. No primary dataset was found in the sources that could be read that measures how often people choose 37 out of 1 to 100, so treat any exact percentage you see attached to 37 as unverified until you can trace it to a dataset.

The 37% rule, with a worked example

The one place where 37 is a mathematical result is the secretary problem, also called the optimal stopping problem. You interview candidates one at a time in random order, you must accept or reject each on the spot, you cannot go back, and you win only if you pick the single best. Weill Cornell Medicine's lecture notes give the analytical solution: observe the first 1/e of the pool without committing, where e is about 2.718, then commit to the first candidate better than everyone seen so far. They credit Thomas Ferguson's 1989 article "Who Solved the Secretary Problem?" for the result.

1/e is about 0.368, so the rule is often rounded to 37 percent. With a pool of exactly 100, the standard formula for the chance of success after rejecting the first r candidates is (r/100) times the sum of 1/(k-1) for k from r+1 to 100. Calculated here, the best r is 37, with a success probability of about 37.1 percent. The same calculation gives these optimal cutoffs for other pool sizes:

CandidatesBest number to skipChance of picking the best
10339.9%
20738.4%
501837.4%
1003737.1%
100036836.8%

Two cautions apply. The rule assumes you are rewarded only for the very best candidate, and it assumes candidates arrive in random order. If second-best is acceptable, or if you can recall earlier options, the optimal strategy changes.

How to test the "random 37" idea yourself

Ask 50 people to name a random number from 1 to 100, record the answers, and compare the counts against what an even spread would give. If everyone chose evenly across 100 options, each number would be expected about 0.5 times in 50 responses, so a value chosen three or four times would stand out. A small informal test will not settle anything, but it shows how lopsided human choices can be, and it keeps the distinction clear between the arithmetic of 37 and the psychology of picking it.

Take this further with Oakgen Agent

Agent can run Python in a sandbox to simulate the problem, produce the table and optionally save the results as a csv or xlsx file.

Looking for a recap of one specific YouTube video? The YouTube Summarizer does that today.

Open this task in Oakgen Agent

The prompt is filled in for you. Nothing is sent until you press send.

Open in Oakgen Agent

Frequently asked questions

Is 37 a prime number?

Yes. Wolfram MathWorld lists 37 as the 12th prime number. You can check it by hand: it is enough to test the primes up to 6 (the square root of 37 is about 6.1), and 2, 3 and 5 do not divide it, so it is divisible only by 1 and itself.

What is the 37% rule?

It is the common name for the optimal strategy in the secretary problem. You look at roughly the first 1/e (about 36.8%) of candidates without choosing anyone, then take the first later candidate who beats everyone you have seen. Weill Cornell Medicine lecture notes describe 1/e as the optimal proportion to observe first. For exactly 100 candidates, the best cutoff works out to 37 under the standard calculation.

Do people really pick 37 when asked for a random number?

MathWorld's entry for 37 says its prominence in requests for a supposedly random two-digit number is a psychological and cultural phenomenon, not a result of its arithmetic. The older, well-documented finding is about single digits: in a 1976 study of 558 passersby on the Yale campus, 28 percent of those asked for a number from 0 to 9 chose 7, according to a University of Michigan statistics textbook that cites the paper.

Why is 1/37 interesting?

37 times 27 equals 999, so 1/37 repeats as 0.027027027... and 37 divides 111 (3 x 37). These are checkable arithmetic facts, and they are unrelated to why people choose the number.

Sources

  1. Wolfram MathWorld, 37 (read 2026-11-19)
  2. Weill Cornell Medicine, L. Skrabanek, The Optimal Stopping Problem, lecture notes, September 24, 2024 (read 2026-11-19)
  3. University of Michigan STAT 216 textbook, 2.6 Activity: A preference for 7? (read 2026-11-19)
  4. Kubovy lab publications list (University of Virginia), citation for Psotka and Kubovy 1976 (read 2026-11-19)
  5. Simon and Primavera, Investigation of the Blue Seven Phenomenon in Elementary and Junior High School Children, Psychological Reports 1972, abstract via ScienceGate (read 2026-11-19)
  6. Wang et al., Number preferences in lotteries, Judgment and Decision Making, Cambridge University Press (read 2026-11-19)
Share