What is the highest IQ ever recorded?
The highest widely cited figure is 228, recorded for Marilyn vos Savant. Here are the real claimants and their numbers, followed by the reason none of them can be ranked against each other.
The question has an answer, several in fact, and then a reason none of them settle it. Both halves are worth having.
The actual figures
| Person | Claimed IQ | Where it comes from |
|---|---|---|
| Marilyn vos Savant | 228 | Stanford-Binet at age ten; listed by Guinness 1986 to 1989 |
| Terence Tao | 220 to 230 | Childhood testing in Miraca Gross’s gifted study |
| Kim Ung-yong | 210 | Childhood testing, listed by Guinness |
| Christopher Langan | 195 to 210 | High-ceiling tests, some untimed and unsupervised |
| William James Sidis | 250 to 300 | Press speculation; no test record exists |
Vos Savant’s 228 is the highest figure with an actual test behind it, and it is the one Guinness listed. Sidis’s numbers are higher and have no test behind them at all.
Why they cannot be ranked
Three different problems, each sufficient on its own.
Most are ratio IQs
Early childhood tests calculated IQ as mental age divided by chronological age, times 100. A ten-year-old performing like a twenty-two-year-old scores 220.
That arithmetic produces enormous numbers for precocious children and it is a different unit of measurement from a modern deviation score. It also does not survive into adulthood, because mental age stops climbing while chronological age does not. The full mechanics are in what IQ stands for, and the Stanford-Binet is where most of these figures originate.
Vos Savant, Tao and Kim all have ratio scores. They are real results on real tests, and they do not convert.
The modern scale does not extend that far
A deviation IQ is a position in a distribution. As IQ percentiles sets out, each step away from 100 corresponds to a rarity.
An IQ of 200 is more than six standard deviations above the mean, a frequency on the order of one in a billion. An IQ of 300 is beyond thirteen standard deviations, describing a rarity vastly exceeding the number of humans who have ever existed.
So a claimed 300 is not an extraordinary measurement. It is a number the scale does not define.
High-ceiling tests are normed differently
Langan’s figures come from tests built specifically to discriminate at the extreme upper tail. Those instruments serve a genuine purpose, but several are untimed, taken unsupervised, and normed on small self-selected samples. The outputs are not comparable to a proctored clinical result.
Why Guinness gave up
Guinness listed a Highest IQ record for years, including vos Savant. The category was retired in 1990, in substantial part because scores from different tests, scales and eras could not be reconciled into a single ranking.
That remains the honest institutional verdict, and nothing since has changed it.
Where the real ceiling sits
Well-normed clinical tests stop discriminating meaningfully somewhere around 145 to 160 depending on the instrument. Beyond that there are simply too few people in the norming sample to separate scores.
Which means that above roughly 160, the correct answer is not a number. It is “higher than this test can measure.”
Our own free IQ test stops at 145, on sixteen items, and says so on the methodology page rather than inventing precision it does not have. A test claiming to place you at 180 is telling you something about the test.
Key takeaways
- The highest figure with a real test behind it is 228, for Marilyn vos Savant, from a childhood Stanford-Binet.
- Most record scores are ratio IQs, a different unit of measurement that does not convert to modern scores.
- Modern deviation IQ does not extend to 200 or beyond in any meaningful sense.
- High-ceiling test results are normed differently and are not comparable to clinical scores.
- Guinness retired the highest IQ category in 1990 because the scores could not be compared.
- Well-normed tests stop discriminating somewhere around 145 to 160.