Bill Gates's IQ The Evidence and an Honest Estimate
No verified IQ test exists for Bill Gates, and the famous 160 is an extrapolation, not a measurement. But unlike most celebrities, he left real cognitive evidence behind. Weighing it carefully, a neutral estimate is most likely 140 to 150. Here is exactly how it is derived. Measure your own real score free.
0 Quick Answer
Bill Gates has never taken a publicly documented, professionally administered IQ test, so no verified score exists. The figure that circulates most often, 160, is not a test result. It is an extrapolation from his near-perfect SAT score, and it ignores an important statistical correction explained below.
Direct answer, stated plainly: Gates is an unusual case, because he left behind two genuinely hard pieces of cognitive evidence that most public figures never produce. He scored 1590 out of 1600 on the pre-1995 SAT, a harder and far less coachable test than today's version that correlated strongly with general ability, and as a Harvard undergraduate he solved an open problem in combinatorics and co-published a result (the pancake-sorting bound, with Christos Papadimitriou) that stood as the best known for roughly thirty years. These are demonstrations of reasoning, not just credentials. Weighing them, while correcting honestly for the imprecision of converting a near-ceiling test score, the most defensible estimate places his IQ most likely in the 140 to 150 range (gifted to very superior), with a roughly 90 percent confidence band of about 132 to 158. The widely cited 160 sits at the optimistic top of that picture rather than at its center. This guide shows exactly how the estimate is derived, where the 160 came from, and why even a high IQ explains only part of what he built.
No. There is no public, professionally administered IQ score for Bill Gates, and that single fact is the honest foundation for everything else here. He has never reported sitting a standardized intelligence test under proctored conditions, and no biographer or interviewer has ever produced one. Every specific IQ number attached to his name is therefore an estimate or an inference rather than a measurement scored against proper norms.
This is the normal state of affairs. Adults are almost never formally tested for IQ, because cognitive testing is usually done in childhood, in clinical settings, or for specific accommodations, none of which applies to a software executive. What makes Gates different from most celebrities is not that he has a tested score, because he does not, but that the public record contains real cognitive data of a different kind: a documented near-ceiling performance on a demanding standardized test, and a published mathematical result. Those let us estimate with more confidence than usual, while still respecting that an estimate is not the same thing as a test, a distinction set out in What IQ Scores Mean.
2 Where the "160 IQ" claim came from
The number you see most often is 160, and unlike a pure invention it has a traceable logic behind it, which is worth understanding precisely because the logic is half right. Gates scored 1590 out of 1600 on the SAT he took in the early 1970s, and the SAT of that era correlated strongly with general cognitive ability. Someone took that near-perfect score, read it through an SAT-to-IQ conversion, and arrived at a figure around 160. The score is real. The conversion is where the trouble begins.
The first problem is that SAT-to-IQ conversions at the very top of the scale disagree wildly with one another. Depending on which table or formula you use, the same 1590 has been mapped to anything from roughly the high 140s to the high 150s, and a few aggressive conversions push higher. When different reasonable methods produce a forty-point spread for the same input, no single output from them deserves to be quoted as a fact. The 160 is simply the high end of that disagreement presented as if it were the answer.
The second and deeper problem is regression to the mean, the statistical correction that the 160 quietly skips. Because the SAT is an excellent but imperfect proxy for IQ, a near-ceiling score is the joint product of very high ability and a favorable performance, so the best estimate of the underlying IQ is pulled somewhat toward the mean rather than read straight off the test. Treating a 1590 as if it equals a 160 IQ assumes the SAT is a perfect measure of intelligence, which it is not. Correcting for this does not erase how exceptional the score is; it just moves the honest estimate from the high 150s down into the 140s, which is exactly what the next sections do.
3 Why a low estimate would be implausible
Neutrality means applying the same scrutiny to low guesses as to high ones, and in Gates's case a below-average or merely-average estimate collapses immediately under the evidence. A near-perfect score on a hard, untimed-feeling power test, survival in Harvard's notoriously brutal Math 55 sequence, and an original combinatorics result published in a peer journal are not things an average reasoner produces. The floor of any honest estimate is high.
So both extremes can be set aside, but not symmetrically. The case against a low number is overwhelming and rests on hard demonstrations of reasoning. The case against the inflated 160 is more subtle: it is not that 160 is impossible, but that it is the optimistic edge of a noisy extrapolation rather than the best central estimate. The honest task that remains is to combine the genuinely strong evidence into a range that is both high, because the evidence is strong, and bounded, because the evidence is still indirect.
4 What we actually know: the SAT 1590 and what it means
The single most informative fact is the SAT score, so it deserves to be read carefully rather than waved at. Gates has stated publicly that he scored 1590 out of a possible 1600, and he was a National Merit Scholar who graduated from the Lakeside School in 1973 before enrolling at Harvard that autumn. In the early 1970s a 1590 was extraordinarily rare, achieved by a tiny fraction of an already-selected pool of college-bound test takers.
Two features of that era's test make the score a better cognitive signal than a modern SAT would be. First, the pre-1995 SAT had a higher ceiling and was less susceptible to preparation and coaching than later versions, so a top score reflected ability more and practice less. Second, formal studies of the relationship between SAT performance and tested intelligence have found a strong correlation at the population level, on the order of the 0.7 to 0.8 range in well-known analyses, which is high for any single proxy. That is why a near-perfect old SAT is treated here as the closest thing to a real cognitive measurement in the entire record, even though it is still not an IQ test. For where the resulting bands sit in the population, see the IQ Score Chart and Gifted IQ Range.
5 The pancake problem: a real demonstration of reasoning
The second hard data point is the one most people have never heard of, and it matters more than any biography of business success. As a Harvard undergraduate, Gates encountered an unsolved problem posed by his professor Harry Lewis in a combinatorics course: given a stack of differently sized pancakes, what is the largest number of spatula flips ever needed to sort them in order? Within days Gates returned with an efficient algorithm, and with Christos Papadimitriou he published the result in 1979 as "Bounds for Sorting by Prefix Reversal."
What makes this so telling is the durability of the work. The upper bound the two of them established stood as the best known solution for roughly thirty years, until a team of researchers improved it decades later. This is the only academic paper Gates ever published, and it is a genuine contribution to theoretical computer science produced by an undergraduate in his spare time. Solving an open research problem is a far stronger signal of fluid and quantitative reasoning than any test score, because it requires generating a novel method rather than recognizing a correct answer, which is the very thing high-end fluid reasoning describes.
This is the piece of evidence that keeps the estimate from regressing too far downward. The regression-to-the-mean correction in section 2 pulls a near-ceiling SAT toward the center because a single test is noisy. But a second, independent demonstration of top-tier reasoning, in a completely different format, reduces that noise. When two unrelated signals both point to the extreme upper range, the case that the underlying ability is genuinely very high gets much stronger, and the honest estimate settles higher within the band rather than at its floor.
6 How you estimate IQ without a real test
If there is no proctored score, can anything responsible be said? Yes, but only as an estimate with honest error bars. Researchers who study the intelligence of people who were never formally tested use historiometric methods, inferring a likely IQ from verifiable indicators rather than from an exam the person never sat. In Gates's case three indicators do most of the work, and they are unusually strong:
Standardized test performance. A near-ceiling score on a test that correlates strongly with general ability, read with the proper statistical correction.
Demonstrated technical reasoning. An original, durable mathematical result and survival in the hardest undergraduate mathematics track.
Educational and strategic record. The schooling completed and the complex, sustained problem solving visible across a technical career.
The crucial caveat is that these estimate a likely region, not a personal score, so the uncertainty for any single individual remains real even when the evidence is good. The principle that keeps the estimate honest is convergence: no single indicator is trustworthy alone, so the goal is to see whether independent signals point to a similar region. Gates is a comparatively favorable case for this method, because his signals are both strong and independent of one another, the test and the published proof drawing on different abilities, which is why his estimate can be tighter than the wide bands appropriate for figures with thinner records.
7 Method 1: the old SAT as a cognitive proxy
The first anchor is the 1590, converted with discipline rather than read off optimistically. Start from how rare the score was: on the pre-1995 test, a 1590 sat very far out in the upper tail of an already-selected population. If you treated the SAT as a perfect intelligence test, that position would imply an IQ near or above 160. But the SAT is not a perfect intelligence test, and the correlation, though strong, is not 1.0.
Applying regression to the mean is the responsible step. Because the test and true ability correlate at roughly 0.7 to 0.8 rather than perfectly, the best estimate of the underlying IQ is the observed standing multiplied by that correlation, which pulls a near-ceiling result down from the high 150s into the mid-to-high 140s. This is not a way of diminishing the achievement; it is the standard correction any psychometrician applies when predicting one measure from another. Read this way, the SAT alone places Gates very comfortably in the gifted range, most plausibly somewhere around the mid-140s as a central figure, with real uncertainty on either side because the conversion at the extreme tail is genuinely imprecise. For how rarity climbs at these levels, see the IQ Rarity Calculator.
8 Method 2: demonstrated technical and mathematical reasoning
The second method leans on what Gates actually did with his mind, which is in some ways more informative than any test. The pancake-sorting result is the centerpiece: producing a novel algorithm for an open problem, and a bound that survived for three decades, is a direct demonstration of high-end fluid and quantitative reasoning. Recognizing a right answer is one thing; generating a method nobody had found is another, and it is the harder one.
Around that sits a pattern of the same kind. He reportedly took Math 55, the legendarily difficult Harvard mathematics sequence that selects hard for raw ability, and he had taught himself to program as a teenager on early time-shared computers, building working software years before that was common. Across his career he remained technically engaged at a deep level, able to review code and architecture rather than only manage. These do not convert into a number, but they corroborate the test in a way that ordinary biographical success cannot, because they are specifically cognitive rather than merely impressive. This is why the estimate does not regress all the way to the floor the SAT correction alone might suggest: a second, independent line of evidence keeps it in the upper part of the gifted range. Generating novel solutions is exactly what the CHC model calls fluid reasoning, one of the most g-loaded abilities a real test measures.
9 Method 3: education and strategic reasoning
A third indicator is the broader record of schooling and decision making, which is consistent with the first two without adding as much precision. Gates was a National Merit Scholar admitted to Harvard, and although he famously left before finishing to start Microsoft, that was an opportunity-cost decision at the dawn of personal computing rather than any reflection of academic limitation. Leaving Harvard to found one of the most valuable companies in history is not a cognitive ceiling; it is a bet that paid off.
The career that followed involved sustained complex problem solving: technical architecture, competitive strategy, and the management of an enormous organization through repeated platform shifts. The honest caveat is that business success is heavily multiply determined. It draws on timing, risk tolerance, work ethic, an enormous team, ruthlessness in competition, and a great deal of luck, none of which an IQ test measures, and early advantages played a role as well. So the strategic record, like the others, argues firmly against a low estimate and is consistent with a high one, while isolating the purely cognitive component less cleanly than the test or the proof. It is best read as supporting evidence that the high estimate from the first two methods is not contradicted by how he actually operated, a relationship explored in IQ and Success.
10 Putting it together: a 90 percent range
Combining the methods, while respecting their limits, produces a defensible estimate rather than a single confident number. The near-ceiling old SAT, corrected for regression to the mean, points to the mid-to-high 140s. The published combinatorics result and the broader technical record independently corroborate top-tier reasoning, which keeps the estimate from sliding to the floor. The educational and strategic record is consistent with all of it. No part of the evidence supports an average score, and the inflated 160 turns out to be the uncorrected reading of the one test rather than a separate confirmation.
Weighing those together, the most defensible estimate places Bill Gates's IQ most likely in the 140 to 150 range, gifted to very superior. Because the inputs are still indirect, and because converting a near-ceiling score carries real imprecision even with two corroborating signals, an honest 90 percent confidence band runs from roughly 132 to 158. Read that band the way a professional reads any score: the interval is the real answer, not a single point. It says he is almost certainly far above the population average and very likely in the gifted range, while the popular 160 sits at the optimistic upper edge rather than the center. This is an estimate built from evidence, not a measurement, which is the difference explained in Reliability & Validity.
11 Why celebrity IQ estimates are unreliable
It helps to step back and see why this whole exercise is error-prone, because the same caution applies to every celebrity IQ number you will ever read, even the better-supported ones. None of these figures is a test result. They are reconstructions from fragments, and even when, as with Gates, the fragments include something solid like an SAT score, the conversion into an IQ number introduces assumptions that can be wrong or chosen to flatter. A real-looking number can still rest on a shaky last step.
There is a recycling effect on top of that. Once a figure like 160 appears in one place, it gets quoted by the next, until the original extrapolation is buried under layers of repetition that read like confirmation. By the time most people meet the number, the regression correction that should have been applied is invisible, and the figure carries the false authority of consensus. The same skepticism is warranted for the numbers attached to Elon Musk, Albert Einstein, Stephen Hawking, and Donald Trump, none of which rest on modern verified tests either. The lesson is not that estimates are worthless, but that a number without its method and its uncertainty is a story, however solid its starting point looked.
12 What IQ does and doesn't explain about his success
Even a perfectly measured IQ would explain only a slice of what Bill Gates achieved, and saying so is part of an honest account. General cognitive ability relates, on average, to learning speed and to performance in cognitively demanding work, as covered in IQ and Success. High ability was clearly necessary for what he did at the technical level. But it was nowhere near sufficient, and treating his IQ as the explanation for Microsoft would badly miss the point.
Building the most valuable software company of its era drew on traits no IQ test captures: an almost unmatched competitive drive, a willingness to bet everything on personal computing before it was obvious, relentless work habits, a shrewd instinct for licensing and platform strategy, the ability to recruit and direct enormous teams, and a large measure of timing and luck in being the right person with the right skills at the start of a revolution. Many people with comparable raw ability never come close to that outcome, and the difference lies mostly outside cognition. Research on achievement supports this: beyond a high-enough threshold, additional IQ points add little to real-world outcomes compared with motivation, temperament, and circumstance. Reading his career primarily through an IQ figure, even a well-estimated one, would point at the wrong explanation.
13 Where a 140 to 150 estimate sits
To make the estimate concrete, here is where the 140 to 150 band falls on the standard scale (mean 100, standard deviation 15), alongside the wider 90 percent interval:
IQ range
Classification
Approx. percentile
Roughly how rare
155 and above
Profoundly gifted (the uncorrected 160 claim)
~99.99th+
~1 in 10,000 and rarer
140–154
Gifted to very superior (most likely estimate)
~99.6th–99.9th
~1 in 261 to 1 in 4,300
130–139
Gifted / very superior (lower band)
~98th–99th
~1 in 44 and rarer
120–129
Superior
~91st–97th
~1 in 11
90–109
Average
~25th–73rd
~1 in 2
The estimate puts him firmly in the gifted region, with the popular 160 sitting just above the top of the honest band rather than inside it. For how these classifications work and how steeply rarity climbs near the top, see the IQ Rarity Calculator and High Average IQ.
14 The honest takeaway
The clean summary is this: nobody can hand you Bill Gates's IQ as a fact, because the test that would produce it does not exist on the public record. What can be offered responsibly is an unusually well-supported estimate, most likely 140 to 150 with a 90 percent band of about 132 to 158, built from a near-perfect old SAT corrected for regression to the mean, a durable published mathematical result, and a long record of technical reasoning, and explicitly not from the uncorrected 160 that circulates online.
That gap between a confident headline and an honest estimate is the whole problem with celebrity IQ numbers, and it is also the reason a real test matters. Gates's case is about as good as outside estimation ever gets, with two strong independent signals, and even here the result is a range several points wide rather than a single figure. The only way to know an actual IQ, yours or anyone's, is to sit a broad, properly normed assessment and read the result with its confidence interval. Estimating a stranger from the outside will always be a guess; measuring yourself directly is the alternative, and it is the one part of this topic you can actually control.
The contrast with a converted SAT score makes clear what a real IQ requires. A genuine score is not a single clever number or a proxy borrowed from another test; it is your standardized standing relative to a defined reference population, built from a broad sample of cognitive tasks and reported with a margin of error. That is why a real score cannot be inferred from one exam taken at seventeen, however impressive, or from a career, however successful.
ACIS measures general cognitive ability the way serious instruments do, by sampling six broad cognitive domains across 20 subtests and combining them into a Full Scale IQ with a confidence interval. It interprets results within a defined adult reference frame and publishes its reliability and validity evidence in the Technical Manual, summarized in Reliability & Validity. The difference between that and a viral IQ chart is the difference between a measurement and a story, the same distinction drawn in Accurate IQ Test and Free vs. Validated IQ Tests.
16 The six domains a real test measures
Where a celebrity estimate collapses everything into one rumored number, a real assessment reports a profile. ACIS is organized around the Cattell-Horn-Carroll (CHC) model and measures six broad domains, which together give the Full Scale IQ its meaning:
Domain
What it captures
Verbal Comprehension
Knowledge, word meaning, and verbal reasoning
Fluid Reasoning
Novel problem solving and pattern detection
Quantitative Reasoning
Numerical reasoning and knowledge
Visual-Spatial
Mental rotation and spatial logic
Working Memory
Holding and manipulating information
Processing Speed
Fast, accurate cognitive throughput
A very strong fluid and quantitative profile, the kind Gates's pancake result would predict, can sit alongside different scores elsewhere, which is exactly the nuance no outside estimate of a public figure can capture. For a deeper treatment, see What an IQ Test Measures and Full Scale IQ.
17 His estimate next to other famous figures
Seeing this estimate in context underlines both its strength and its limits. Most famous IQ numbers rest on nothing but legend. Albert Einstein is routinely assigned a 160, yet he never took a modern IQ test, so that figure is itself a popular invention, as discussed in Albert Einstein's IQ. The same is true for the numbers attached to Leonardo da Vinci and Nikola Tesla, and for the wide ranges around Elon Musk and Taylor Swift.
Gates is a comparatively strong case within that weak field, because he left an actual standardized score and a published proof rather than only anecdotes. That is why his band can be set higher and tighter than the estimates for figures whose records are thinner. But the key point is that even his estimate, the best-supported of the group, is still a range derived from evidence rather than a measured score, and it still corrects a number that the internet treats as fact. The honest comparison is not "who is smarter" based on invented figures, but a recognition that all of these numbers, strong and weak alike, share the same final weakness: none is a real, modern, normed test.
The estimate on this page is offered in the opposite spirit from a viral number: with its method shown, its central correction made explicit, and its uncertainty stated. That is the only kind of celebrity IQ figure worth anything, and even at its best it remains a reasoned estimate rather than a measurement. It is precisely why a real, comparable score is worth more than any figure a chart can assign from the outside, and why the most useful thing this page can point you toward is not his number but your own.
18 Common myths about Bill Gates's IQ
"His IQ is 160." That number is an uncorrected reading of his SAT, not a test result. Converting a near-ceiling score without regression to the mean inflates it; the corrected estimate lands lower, in the 140s.
"The SAT score proves a 160 IQ exactly." SAT-to-IQ conversions at the top of the scale disagree by dozens of points, so no single output from them is a fact, only a range.
"He dropped out, so the degree question counts against him." He left Harvard to found Microsoft at the start of personal computing, an opportunity-cost choice, not an academic limitation.
"A high IQ is why Microsoft succeeded." Ability was necessary but far from sufficient; drive, timing, strategy, and luck explain most of the outcome.
"An estimate this confident settles it." Even his unusually well-supported estimate is a range from evidence, not a measurement, and carries real uncertainty.
The Gates case is a useful example of a general truth, including its best-case version. You can form a rough, range-shaped estimate of almost anyone from solid facts, chiefly their education and the cognitive demands of work they have demonstrably done. When the facts happen to include hard cognitive data, like a standardized test score and an original technical result, the estimate gets tighter and more confident than usual, as it does here. What you still cannot do is produce a precise point score, because even good proxies carry conversion error and noise.
His case also shows the right discipline when the data is strong: apply the standard corrections, here regression to the mean, rather than reading the most flattering number straight off the page, and let independent signals raise or lower the estimate within an honest band. That is why responsible sources give ranges and corrections even for well-documented figures, and why a single number presented without method or uncertainty is a red flag regardless of how solid its starting point looks. The same discipline that produces an honest estimate of a public figure is what a real test applies to you directly, only with far more data and proper norms behind it, from How IQ Scores Are Normed to What IQ Scores Mean.
20 Bottom line
Bill Gates's IQ is unknown in the only sense that counts: there is no verified test score. But his record is unusually rich, so the evidence-based estimate is both high and reasonably tight: most likely 140 to 150, with a 90 percent confidence band of about 132 to 158, derived from a near-perfect old SAT corrected for regression to the mean, a durable published combinatorics result, and a long record of deep technical reasoning, and explicitly not from the uncorrected viral 160. He is almost certainly far above average and very likely gifted, and beyond that the evidence does not reach.
If a precise number is what you actually want, the only place to get one is a real test taken under proper conditions, not a chart about someone else. You cannot test a public figure from your screen, but you can measure yourself in about an hour and read a real Full Scale IQ with a genuine confidence interval. That is the one number in this entire conversation that can be more than a guess.
