The Highest IQ Ever Recorded: Why There Is No Answer

The record that no longer exists

From the mid-1980s until 1989, the Guinness Book of World Records listed Marilyn vos Savant under “Highest IQ”, with a score of 228. In 1990 the category was removed from every subsequent edition.

The reason Guinness gave is the most important sentence in this entire subject: IQ tests were too unreliable to designate a single record holder. Not that vos Savant's score was wrong. That the question could not be answered by the instruments available.

Where 228 came from

In September 1956, aged ten, Marilyn vos Savant sat the 1937 Stanford–Binet, Second Revision. The test assessed her mental age at twenty-two years and ten months. Applying the ratio formula — mental age divided by chronological age, multiplied by 100 — produces approximately 228.

Two facts about that calculation are decisive, and both come from the test's own documentation rather than from later critics.

1. The 1937 Stanford–Binet manual's norms do not permit IQs above 170 at any age. A score of 228 lies far outside the range the test was built to report.
2. Terman and Merrill wrote that beyond age fifteen, the mental ages “are entirely artificial and are to be thought of as simply numerical scores.” A mental age of twenty-two years and ten months is therefore not a measurement of anything. It is a number produced by extrapolating a scale past the point where its authors said it stopped meaning anything.

Attempts to convert her result onto a modern deviation scale generally land somewhere between 180 and 190 — still extraordinary, and roughly forty points below the number that made her famous.

Ratio IQ is why childhood scores look enormous

The whole phenomenon of implausibly high IQs traces to one obsolete piece of arithmetic.

A ratio IQ divides mental age by chronological age. For a young child, the denominator is tiny, so a modest absolute advantage produces a huge quotient. A six-year-old performing like a twelve-year-old scores 200. A thirty-year-old performing like a sixty-year-old scores nothing at all, because mental age stops tracking birthdays somewhere in adolescence — which is precisely the objection Terman and Merrill themselves raised.

David Wechsler replaced the formula in 1939 with the deviation IQ: your score is your position within the distribution of people your own age, with the mean set at 100 and one standard deviation at 15. Every modern test uses it. The two quantities share a name and nothing else.

The arithmetic of very high scores

Set aside the history and ask what a modern score of 228 would have to mean.

On the standard scale, 228 is more than eight and a half standard deviations above the mean. The implied rarity is on the order of one person in 10^17. Roughly 10^11 human beings have ever lived. For the claim to be coherent, the human species would need to be about a million times larger than it has ever been.

This is why serious clinical instruments stop reporting somewhere near 160, and why any specific three-digit number above 200 should be treated as a claim about arithmetic rather than a claim about a person.

The modern claims of 200, 250, 276

New record holders are announced periodically, with numbers that climb each time. They share a common feature.

These scores come from unstandardised “high-range” tests — untimed puzzle sets, often taken at home, marketed to and by high-IQ societies. They have no representative standardisation sample, because assembling one at four or five standard deviations above the mean would require testing millions of people. Their percentile claims are extrapolations from small, self-selected groups of people who sought out a very hard test. Mainstream psychometrics does not accept them, and neither does Guinness, which is why the record stays retired.

The same applies to the historical figures. William James Sidis is routinely credited with an IQ between 250 and 300. No documented test score for him exists. Cox's 1926 estimates for people such as Goethe and Newton are ratio IQs reconstructed from biographical records, and their subjects were all born between 1450 and 1850 — a fact that also excludes, for example, Einstein.

So who does have the highest IQ?

Nobody knows, and the honest position is that the question is not well formed.

To identify the single most intelligent person alive you would need an instrument that discriminates reliably among the top one in a billion. No such instrument exists, and constructing one would require a norm sample larger than the population you were trying to rank. The problem is not that the measurement is difficult. It is that the measurement is self-defeating.

What can be said is narrower. Some people reason exceptionally well. Enough of them exist that they are not individually remarkable. And whatever separates a Newton or a Ramanujan from the many thousands of people with comparable measured ability, no test has ever caught it — which is the finding Lewis Terman's own lifelong study of high-IQ children accidentally established.

Myths and facts