Short answer: 228 is best known as Marilyn vos Savant's childhood ratio IQ, the figure Guinness listed under "Highest IQ" in the late 1980s. It came from dividing mental age by age, not from a modern bell-curve scale. As a deviation IQ (SD 15), 228 would sit 8.5 standard deviations above average, about 1 in 141 quadrillion, a level no test can verify.
An IQ this high sits beyond the right edge of the curve: no normed test can reach it.
At a glance
| Percentile | Above the 99.999999999999999th (theoretical) |
|---|---|
| Rarity | About 1 in 140.7 quadrillion on an SD 15 curve (z = 8.53); theoretical only |
| Wechsler classification | Very Superior (WAIS-IV) / Extremely High (WAIS-5) by label, but above the WAIS ceiling of 160 |
| Stanford-Binet 5 classification | Very gifted or highly advanced by label, but beyond the SB5's 160 ceiling |
| On an SD 16 scale | 237 |
| On an SD 24 (Cattell) scale | 305 |
| Mensa eligible? | Any verified score this high clears Mensa's 130 line, but modern tests cannot report it |
| Best-known holder | Marilyn vos Savant (1937 Stanford-Binet, taken at age 10) |
Where 228 comes from
Marilyn vos Savant, born Marilyn Mach in St. Louis on August 11, 1946, took the 1937 Stanford-Binet, Second Revision, at age 10. By her account the test was given in September 1956 and placed her mental age at 22 years and 10 months. Divide 22.83 by 10, multiply by 100, and the result is 228.
Guinness listed that figure under "Highest IQ" from 1985 to 1989 and inducted her into its Hall of Fame in 1988. In 1990 it retired the category, concluding that IQ tests were too unreliable to name a single record holder. The full story of her scores is on our Marilyn vos Savant IQ profile.
Ratio IQ versus deviation IQ
The two systems measure different things. A ratio IQ compares a child's performance with the typical performance of older or younger children: mental age over chronological age, times 100. William Stern proposed the quotient in 1912; Lewis Terman's 1916 Stanford-Binet made it famous.
A deviation IQ ranks a person against people of the same age, on a bell curve with mean 100 and standard deviation 15. Wechsler tests use it, and the Stanford-Binet adopted it with its 1960 revision. The 1937 forms that produced the 228 were the last to report ratio scores.
The difference matters most at the extremes. A bright young child can outpace age norms by years, so ratio scores inflate quickly; the original Stanford-Binet could yield IQs above 200. A deviation scale has no such stretch. A ratio 228 therefore cannot be read as a deviation 228, and no official conversion exists for a single childhood result.
What 228 would mean on a bell curve
Treated as a modern score, 228 sits (228 − 100) ÷ 15 = 8.53 standard deviations above the mean. The normal curve puts the share of people at that level near 1 in 140,656,629,424,771,088, roughly 1 in 141 quadrillion.
| IQ | z-score | Expected rarity |
|---|---|---|
| 160 | 4.0 | 1 in 31,574 |
| 200 | 6.7 | 1 in 76 billion |
| 220 | 8.0 | 1 in 1.6 quadrillion |
| 225 | 8.3 | 1 in 25 quadrillion |
| 228 | 8.5 | 1 in 141 quadrillion |
| 250 | 10.0 | 1 in 1.3 × 10^23 |
With 8.2 billion people alive, that rarity implies one such person per roughly 17 million Earths. The same curve predicts that the rarest person alive scores near 195. Rarity at every level is charted on our IQ rarity chart.
Why modern tests stop at 160
Deviation scores rest on norm samples. The WAIS-IV, normed on 2,200 people, reports full-scale scores from 40 to 160. The Stanford-Binet 5 ends at the same point. A score of 228 would require norms containing people 8.5 standard deviations above average, and no sample on Earth could supply them.
Precision also thins long before the ceiling. Psychometricians regard reports far above 160 as dubious, because error grows as scores climb. That is the logic Guinness followed when it closed the category. The same arithmetic, pushed further, explains why an IQ of 800 is impossible.
The same performance also changes number by scale: 228 on an SD 15 curve corresponds to 237 with SD 16 and 305 with SD 24. See our guide to IQ classification.
Her second score: the Mega Test's 186
Vos Savant's other widely reported number came from the Mega Test, an unsupervised take-home test built by Ronald Hoeflin and taken in the mid-1980s. Hoeflin reported a raw score of 46 out of 48, which he converted to an IQ of 186 on a scale with standard deviation 16.
Professional psychologists criticized the test as improperly designed and scored, describing its method as "nothing short of number pulverization." Neither the 228 nor the 186 comes from a modern, professionally normed adult test.
What the 228 does and does not show
It shows a ten-year-old who performed at an adult level on a demanding test. Dean Keith Simonton, writing in Nautilus in 2018, noted that she reportedly got a perfect score, which means the result reflects the top of the test, not a measured limit of ability.
It does not establish an exact adult rank, and it does not make her the proven highest IQ in history. Her public record rests on other work: the "Ask Marilyn" column in Parade since 1986, including the September 9, 1990 column on the Monty Hall problem. For other contested extremes, read the highest IQ ever recorded.
Getting a score you can trust
Today a credible IQ comes from a psychologist administering a normed test and reporting a confidence interval of roughly five points each way. Mensa's line is 130, the top 2 percent; our Mensa admission guide explains the accepted tests. Our free IQ test is a practice screener, not a clinical measure; see whether online IQ tests are accurate. For the realistic top of the range, read what a top 1 percent IQ means.
Frequently asked questions
Who has an IQ of 228?
The figure is tied to Marilyn vos Savant. She took the 1937 Stanford-Binet at age 10, which placed her mental age at 22 years and 10 months, a ratio IQ of 228. Guinness listed the number under Highest IQ from 1985 to 1989.
Is an IQ of 228 possible?
As a childhood ratio IQ, yes: ratio formulas could exceed 200 for young children who far outpaced age norms. As a modern deviation IQ, no test can produce it. Adult tests such as the WAIS-IV stop at 160, and 228 would require 8.5 standard deviations above the mean.
What is the difference between ratio IQ and deviation IQ?
Ratio IQ divides mental age by chronological age and multiplies by 100, comparing a child with other age groups. Deviation IQ ranks a person against same-age peers on a bell curve with mean 100 and SD 15. The Stanford-Binet switched to deviation scores in 1960.
Is Marilyn vos Savant still the record holder for highest IQ?
There is no current record. Guinness retired the Highest IQ category in 1990, concluding that IQ tests were too unreliable to designate a single record holder. Her 228 remains the best-known figure from the era when the category existed.
What percentile is an IQ of 228?
On a deviation scale it would exceed the 99.999999999999999th percentile, about 1 in 141 quadrillion. That figure is purely theoretical. It describes a perfect bell curve, not anything a test has measured, and it applies poorly to a ratio score from 1956.
Sources
- Marilyn vos Savant — 1937 Stanford-Binet at age 10; mental age 22 years 10 months; Guinness 1985-1989; category retired 1990; Mega Test 186
- Simonton, D. K. (2018). Your IQ Matters Less Than You Think. Nautilus — Reported perfect score on the Stanford-Binet at age 10
- Ratio and deviation test scores (TAG PDX) — 1937 forms were the last ratio edition; ratio scores could exceed 200
- Intelligence quotient — Stern's 1912 quotient; scores much above 160 considered dubious
- Canivez, G. L. (2010). Review of the WAIS-IV. Mental Measurements Yearbook — WAIS-IV range 40-160; norm sample 2,200
- UN: World population 8.2 billion in 2024 — Population base for rarity comparison
Editorial standard: IQ figures for historical figures are retrospective estimates and are labeled as such; scores for living people appear only when the person disclosed them or a reputable outlet reported them. Corrections: contact the editors.