Average IQ of Harvard Students: About 130, Not 145

The average IQ of Harvard students is about 130, give or take a few points, on the standard scale where 100 is average and the standard deviation, the typical distance of a score from that average, is 15. No one has published a measured average for the whole student body. The estimate rests on two routes that agree: a Wechsler IQ study of 184 Harvard students, whose mean was 129.4, and Harvard's published SAT profile, which implies about 129 to 132 once regression to the mean is allowed for. That is roughly the 97th to 98th percentile of the general population. Claims of 145 or more have no measured source.

We found no IQ figure published by Harvard. What it does publish every year is the SAT profile of its entering class. This article walks through both routes, corrects the common mistakes, and then sets Harvard beside Yale, Stanford, MIT and Cornell, and looks at what Harvard Law School and Harvard Business School publish.

What Harvard's own SAT data show

Colleges report admissions statistics in a standard format called the Common Data Set. Harvard's edition for 2025-26, covering the Class of 2029 that entered in fall 2025, reports these figures:

• 47,893 applicants, 2,003 admitted and 1,675 enrolled
• 79% of enrolled first-year students submitted SAT scores and 27% submitted ACT scores
• SAT Reading and Writing: 720 at the 25th percentile of the class, 750 at the median, 770 at the 75th
• SAT Math: 730, 770 and 790 at the same three points
• 84% of SAT submitters scored 700 or more in Reading and Writing, and 85.6% scored 700 or more in Math

The year before, the medians were higher, at 760 and 790, but only 54% of that class submitted an SAT, so those figures described the students who chose to send scores. The class did not get weaker in 2025-26; the group of submitters widened from 54% to 79%. Harvard does not report a combined SAT median, and adding the two section medians together does not produce one, so we work with the sections. (Yale, Stanford, Cornell and MIT all do report a composite; see the comparison below.)

On the College Board's current percentile table, a 750 in Reading and Writing ranks at the 98th percentile of recent test takers, and a 770 in Math at the 97th. Against a nationally representative sample of US high-school students, both are at the 99th. The second comparison is the relevant one, because far from every teenager takes the SAT: the 2,004,965 test takers in the Class of 2025 are only part of their age group.

Why the SAT tracks IQ, and why the match is not exact

The link between the SAT and IQ comes from a 2004 study by Meredith Frey and Douglas Detterman in Psychological Science. Using a national survey of 917 young Americans who had taken both the SAT and the military's aptitude battery, they found the SAT correlated .82 with a general cognitive ability factor. In a second sample of 103 university students (104 were tested and one outlier was dropped), SAT scores correlated .48 with Raven's Advanced Progressive Matrices, rising to .72 once corrected for the narrow range of ability at a selective university. Our explainer on SAT and ACT score to IQ conversion covers both studies in detail.

A correlation of .82 is strong, but not perfect, and that has a precise consequence. When one measure is used to predict another, the prediction sits closer to the average than the score it was built from. This is regression to the mean. In standard-deviation units the best single guess is the SAT position multiplied by the correlation, so an exceptional SAT result predicts a very good, but less exceptional, IQ.

The arithmetic, step by step

• Start with the median Harvard student's section scores, each at about the 99th percentile nationally. The College Board shows higher scores only as "99+", so the true figure is somewhere between the 99th and the very top
• The 99th percentile is 2.33 standard deviations above the average, and the 99.5th is 2.58
• Multiply by the correlation of .82: 2.33 becomes 1.91, and 2.58 becomes 2.11
• Convert to the IQ scale, 100 plus 15 points per standard deviation: 1.91 gives about 129 and 2.11 about 132
• Frey and Detterman's own equation reports a standard error of prediction of 5.94 points, but it was fitted on a sample of 917 young people who had taken the SAT, a narrower slice of ability than the whole population. With the full 15-point spread and a correlation of .82 the standard error is 15 x sqrt(1 – .82 squared), about 8.6, so one student's predicted IQ carries a 95% margin of about 17 points either way (our calculation, assuming a bivariate normal relationship)

So the SAT route puts the median Harvard student at about 129 to 132. For an individual the range is wide: with a 17-point margin, a student with a 99th-percentile SAT could plausibly sit anywhere from the low 110s to the high 140s. You can check the percentile side of this on our IQ percentile calculator: 129 sits at about the 97th percentile, 132 at about the 98th.

The study that tested Harvard students directly

One published study did give Harvard students an actual IQ test. Shelley Carson, Jordan Peterson and Daniel Higgins, writing in the Creativity Research Journal in 2005, tested 184 Harvard graduate and undergraduate students with the Wechsler Adult Intelligence Scale-Revised. The paper is behind a paywall, so we read its figures from a 2024 meta-analysis by Uttl and colleagues that lists the study: a mean IQ of 129.4 and a standard deviation of 10.9.

That figure lines up closely with the SAT estimate, and a standard deviation near 11 is itself informative. If scores in the group were roughly normally distributed, about two-thirds of the students would fall between about 118 and 140. Harvard is a group of very able people, not a group of identical ones.

• The sample mixed graduate and undergraduate students, and nothing in the published figures shows it was drawn at random from the student body
• We could not check whether the test was a full or a short form, because the paper is paywalled
• The test's norms dated from about 1980, and older norms flatter later test takers, a drift our article on why IQ norms expire explains. The Uttl meta-analysis subtracts 0.3 IQ points for each year between the test's 1980 standardisation and the 2003 testing, which takes the Harvard mean from 129.4 to about 122.5. That adjustment is a rule applied to every sample, not a re-test
• A 2003 Harvard Gazette report on the same research gave 130 as the average IQ of Harvard students in a parenthesis. It counts 182 students, almost certainly the same sample, so it is not a second study

Average IQ of students at Yale, Stanford, MIT, Cornell and other elite universities

No study since 2020 that we could find has measured IQ in Ivy League or other elite students, so for Harvard's peers the evidence is again admissions data. These are the 25th, 50th and 75th percentile SAT scores of the entering class in each school's own 2025-26 Common Data Set, with the share of the class whose scores are included:

• Harvard: no composite reported; Reading and Writing 720 / 750 / 770 and Math 730 / 770 / 790; 79% submitted
• Yale: composite 1480 / 1540 / 1560; 67% submitted
• Stanford: composite 1520 / 1550 / 1570; 56% submitted
• MIT: composite 1520 / 1550 / 1570; 83% submitted
• Cornell: composite 1490 / 1530 / 1550; about 43% submitted (our division of the counts in its data set)

We could not retrieve Princeton's or Columbia's data sets (their sites refused our requests) or the 2025-26 files for Brown, Dartmouth and Penn, so we leave those schools out rather than quote a search summary. The medians that were available all sit within about 20 points of one another, and the College Board lists every total from 1520 up as "99+" on its nationally representative table. A median in that band converts, by the same arithmetic as above, to an IQ-style estimate of at least 129, about 132 if it is at the 99.5th percentile and about 138 if it is at the 99.9th. The published tables do not say how far inside "99+" a median sits, and a test-optional school's median describes only the students who sent scores, so none of this ranks one school above another. For the lower end of the range, see our look at average IQ at MIT, Oxford and Yale.

Average IQ of Harvard Law, Business and Medical School students

Harvard's graduate schools publish admissions statistics, not IQs, and the tests are different ones:

• Harvard Law School J.D., Class of 2029: LSAT 172 / 174 / 176 at the 25th, 50th and 75th percentile of the class, and undergraduate GPA 3.91 / 3.97 / 4.00; 9,061 applications and 561 enrolled
• Harvard Business School MBA, Class of 2027: GMAT Focus median 685 (middle 80% from 645 to 735) and GMAT 10th edition median 730 (690 to 770); 44% of the class submitted the GRE, 34% the GMAT Focus and 28% the older GMAT

We did not read a published average IQ for any of these classes, and we could not retrieve Harvard Medical School's figures. These applicants are already graduates, and tests such as the LSAT rank them against other law-school applicants rather than the whole population, so converting a median LSAT to an IQ is looser than the SAT arithmetic above; our guide to GRE and LSAT scores to IQ explains the method and its limits.

Where the 145 figure comes from

The claim that the average Harvard student has an IQ of 145 appears across forums and blogs, sometimes attributed to well-known academics. We found no measured source for it. The versions we found cite no source or convert a high SAT average straight into the same percentile of IQ, which skips the regression step shown above. An IQ of 145 is the 99.87th percentile, about one person in 741. Only a small fraction of any entering class, even at Harvard, would be expected to score at that level. Across all undergraduates, newer data put the average far lower: the 2024 Uttl meta-analysis of 106 samples and 9,902 students puts it at about 102, falling roughly 0.2 points a year. See also our guide to the average IQ of college students.

Admissions data point the same way. Test scores are one input among many. In an analysis of six Harvard classes, those graduating from 2014 to 2019 (applications from the fall of 2009 to 2014), drawn from data released in the Students for Fair Admissions lawsuit, Peter Arcidiacono, Josh Kinsler and Tyler Ransom found that 29% of admitted domestic students were recruited athletes, legacies, applicants on the dean's interest list, or children of faculty and staff. Recruited athletes made up under 1% of applicants and over 10% of admits. Arcidiacono was an expert witness for the plaintiffs, but the figures come from the applicant data Harvard produced for the case.

What this means for your own score

• An IQ around 130 puts someone in the same range as the typical Harvard student on these estimates, and that is about 1 person in 44 in the general population, as our piece on how rare a high IQ score is explains. Whether 130 counts as genius is a separate question that our guide to IQ 130 takes up
• Admission depends on grades, coursework, recommendations, activities and circumstance as well as ability, and our article on whether IQ predicts school success covers how much of academic achievement a score accounts for
• A single SAT or IQ score is an estimate with a margin, not a fixed property, which is why our note on the margin of error on an IQ score matters here too

The fair answer to "what is the average IQ of Harvard students?" is about 130, give or take a few points, on a mean-100 scale. It rests on two routes, a direct Wechsler IQ study and the published SAT profile, and they agree. The higher figures skip a step of arithmetic or cite nothing at all. If you would like a number for yourself, our IQ test reports a score with its percentile, and our overview of average IQ by education level shows how the figure compares across degrees.

Common questions