The Mensa IQ Score Requirement Is a Percentile, Not a Number

Ask what IQ you need for Mensa and you will be told 130, or 132, or 148, usually with no hint that these are the same requirement expressed three different ways. Mensa does not have an IQ threshold. It has a percentile threshold, and every number attached to it is what that percentile turns into once you name a test.

The actual rule

Mensa admits people who score at or above the 98th percentile on an approved standardised intelligence test — the top two per cent of the general population, as represented by that test's reference sample. The society was founded in Oxford in 1946 by Roland Berrill and Lancelot Ware, and the top-two-per-cent criterion has been the entry rule throughout. Everything else — which tests count, what figure each one requires, how old a score may be — is administration built on top of that one sentence.

The distinction is not pedantry, because a percentile and a score are different kinds of thing. A percentile is a position in a reference sample. A score is that position printed on whatever scale the test happens to use. Move to a test with a different scale and the position stays exactly where it was while the printed number moves.

What the 98th percentile is worth in points

Modern tests are scored so that the average is 100 and the spread is fixed by a chosen standard deviation — a measure of how far apart people's results fall. The 98th percentile sits about 2.05 standard deviations above the mean. Multiply that by the standard deviation in use and add 100:

• Standard deviation 15 — the Wechsler scales and most contemporary tests. The 98th percentile falls at roughly 131.
• Standard deviation 16 — some Stanford-Binet forms and a number of older instruments. Roughly 133.
• Standard deviation 24 — the Cattell scale, long used by British Mensa for its own supervised test. Roughly 149.
• Ratio IQ — the obsolete mental-age formula. It has no fixed standard deviation at all, which is one reason no serious admissions process uses it.

Now set those against the figures in circulation. 130 is exactly two standard deviations on the SD-15 scale, 132 is exactly two on SD-16, and 148 is exactly two on SD-24. Two standard deviations is the 97.7th percentile, not the 98th. The folklore has quietly rounded the threshold down: the difference between the top 2 per cent and the top 2.3 per cent is small, but it is being introduced by people quoting an organisation that chose the phrase "top two per cent" on purpose.

Two ways in, and one that does not count

• Prior evidence — a qualifying result from an approved test that was administered under controlled conditions by someone qualified to give it, submitted for review.
• Mensa's own supervised admission test, sat at a scheduled session under proctored conditions.
• Neither route accepts an unsupervised online score. Mensa's at-home or practice test is offered as an indicator of whether the supervised test is worth booking, not as a qualification in itself.

The reason is not gatekeeping for its own sake. A score means what it claims to mean only if the conditions match the ones the norms were collected under: the same time limits, the same instructions, no interruptions, no second attempt, nobody else in the room. Remove that and the result is still a number, but it is no longer a percentile against the reference sample — which is the only thing Mensa is asking for. Our note on what separates a well-built online test from a bad one works through the same problem from the other end.

The cutoff is administrative, not natural

A threshold has to be somewhere, and the top two per cent is as defensible a line as any. What it is not is a boundary between two kinds of person. Every score carries measurement error, and test manuals publish a figure — the standard error of measurement — describing how far a result bounces around a person's true standing between sittings. On a Wechsler-type scale that error is large enough that a reported 129 and a reported 132 are not reliably different results. One of them is admitted and the other is not, and the honest description of that difference is that it is administrative.

The caution runs in the other direction too. Someone who qualifies has produced one estimate above a line on one day. That is a real result, and it is still an estimate with a range around it.

What membership does and does not tell you

• It tells you that one supervised test placed someone in the top two per cent of a reference sample, on a stated date.
• It does not tell you which abilities were strong. A qualifying total can be reached by very different profiles across the subtests.
• It does not predict any individual outcome. Score-to-outcome relationships in education and work are group averages with wide spread around them.
• It does not expire in any psychometric sense, but the norms behind it do — a score from decades ago was measured against a sample that no longer represents anyone.
• And roughly one adult in fifty meets the criterion by construction, which is a great many people. Mensa's membership is a small fraction of that, because qualifying and joining a society are unrelated decisions.

The useful takeaway is not the number. It is the habit the number should teach: when someone quotes a threshold, ask which scale it is on, and if they cannot say, they are quoting folklore. That applies to Mensa, to the figures attached to famous names, and to any score handed to you without its scale.

Common questions