Updated : 1 day ago
View Complete ArticleChecked 29 September 2026. Percentiles are GMAC's concordance "Published Aug. 2026" (data 1 Jul 2021 to 30 Jun 2026) and the mba.com widget. GMAC updates them every year in the third quarter; the next table will move these numbers slightly.
A GMAT total on the current edition runs from 205 to 805, built from three sections that each score 60 to 90 and count equally: Quantitative Reasoning, Verbal Reasoning and Data Insights. The number itself tells a business school little until it is set against everyone else who sat the exam, which is what the percentile does. GMAC publishes that ranking in two places: a concordance PDF with one decimal, and a widget on mba.com that rounds to whole numbers. Both now carry data from July 2021 to June 2026, and both put a 645 at the 86th percentile (85.6 on the PDF). Two of GMAC's own articles still print older readings: the 87th, which was the July 2025 table rounded, and "the 90th" for 645-655. This guide gives the full current table, the section tables that behave very differently from the total, the link to old 10th Edition scores, and a calculator.
GMAC, mba.com "Understanding your score" widget, data period 2021-2026; support.mba.com "How is the exam scored?".
Where other pages get this wrong
On GMAC's current table a 645 is the 85.6th percentile (86 rounded) and a 655 is the 89.6th (90 rounded). The 87th is last year's figure, and the range 645-655 is not uniformly "the 90th".
Commonly published instead: GMAC's own article on average scores at top programmes (10 Jun 2026): "As of 2026, a 645+ GMAT score means placing in the 87th percentile". GMAC's getting-started article: "a score above 645-655 puts you into the 90th percentile".
The concordance "Published Aug. 2026" gives 645 = 85.6% and 655 = 89.6%; the mba.com widget rounds them to 86 and 90. The 87th matches the July 2025 table (645 = 86.7, rounded), which was still current when the article appeared on 10 Jun 2026; it is stale under the Aug 2026 table. 645 was 87.7 in Aug 2024. Quote the table, not the article, and say which edition you mean.
Every total on this edition ends in 5. The first percentile column is the concordance PDF (one decimal); the second is the same data rounded on mba.com. Where the two seem to differ by a point, it is rounding, not a different dataset.
| Total score | Percentile, Aug 2026 table (official) | mba.com widget (official) |
|---|---|---|
| 805 | 100.0 | 100 |
| 795 | 100.0 | 100 |
| 785 | 100.0 | 100 |
| 775 | 100.0 | 100 |
| 765 | 99.9 | 100 |
| 755 | 99.8 | 100 |
| 745 | 99.6 | 100 |
| 735 | 99.4 | 99 |
| 725 | 98.9 | 99 |
| 715 | 98.6 | 99 |
| 705 | 97.5 | 98 |
| 695 | 96.9 | 97 |
| 685 | 95.1 | 95 |
| 675 | 94.0 | 94 |
| 665 | 91.2 | 91 |
| 655 | 89.6 | 90 |
| 645 | 85.6 | 86 |
| 635 | 80.9 | 81 |
| 625 | 78.2 | 78 |
| 615 | 75.4 | 75 |
| 605 | 69.4 | 69 |
| 595 | 66.3 | 66 |
| 585 | 59.9 | 60 |
| 575 | 56.7 | 57 |
| 565 | 50.4 | 50 |
| 555 | 47.4 | 47 |
| 545 | 41.7 | 42 |
| 535 | 39.1 | 39 |
| 525 | 34.0 | 34 |
| 515 | 31.6 | 32 |
| 505 | 27.3 | 27 |
| 495 | 25.4 | 25 |
| 485 | 21.7 | 22 |
| 475 | 20.1 | 20 |
| 465 | 17.1 | 17 |
| 455 | 15.7 | 16 |
| 445 | 13.2 | 13 |
| 435 | 12.1 | 12 |
| 425 | 10.2 | 10 |
| 415 | 9.3 | 9 |
| 405 | 7.7 | 8 |
Below 405 the table continues in the same way down to 205 (0.2 on the PDF, 0 on the widget); the calculator below reads those rows too. Two features of the table matter in practice. First, the top is flat: everything from 745 up rounds to the 100th percentile on the widget, and 775 to 805 are 100.0 even on the PDF, so above about 735 extra points buy almost no extra rank. Second, the middle is steep: between 585 and 655 each 10-point step moves you three to seven percentile points, because test takers are relatively dense in that part of the distribution.
GMAC recomputes percentiles every year over a rolling five-year window, and the score report you already hold is updated automatically. GMAC's help centre is explicit that "your total score remains unchanged" while the ranking attached to it moves. Over the three recent editions, each middle and upper-middle score in the table below has lost rank, while the very top (775 to 805) stays at 100.0. GMAC's stated reasons are "population shifts, graduate management education trends, and other external forces"; it does not say more simply that recent test takers score higher.
| Focus total | Aug 2024 table (data 2019-2024) | Jul 2025 table (data 2020-2025) | Aug 2026 table (data 2021-2026) | Change 2024 to 2026 (ours) |
|---|---|---|---|---|
| 705 | 98.2 | 98.0 | 97.5 | -0.7 |
| 655 | 91.3 | 90.5 | 89.6 | -1.7 |
| 645 | 87.7 | 86.7 | 85.6 | -2.1 |
| 615 | 77.7 | 76.4 | 75.4 | -2.3 |
| 565 | 52.1 | 50.9 | 50.4 | -1.7 |
GMAC concordance tables, total score. 565 (52.1, 50.9, 50.4) is left off the chart to keep the scale readable. Change column is our subtraction.
Two consequences for an applicant. A school that writes its floor as a percentile rather than a score, as IIM Calcutta does for its one-year MBAEx, can move the score you need from one cycle to the next without changing a word of its policy. And an old article that says "645 is the 88th percentile" may have been right when written; the fix is to check the date on the table it quotes. GMAC's school resources page links the current table as an "Aug2026" file, but an older GMAC article still links a copy named "Jul2025" whose contents also say "Published Aug. 2026", so cite the "Published" line, not the file name.
The three section scales look identical, 60 to 90, but they are not interchangeable. Verbal scores bunch tightly (standard deviation 4.89), Quant spreads widely (6.52), and Data Insights sits lowest on average (mean 75.14). The result: an 80 is the 83rd percentile in Data Insights, the 64th in Quant and only the 54th in Verbal. When you decide what to work on for a retake, compare section percentiles, not section scores.
| Section score | Quant percentile | Verbal percentile | Data Insights percentile |
|---|---|---|---|
| 90 | 100 | 100 | 100 |
| 88 | 95 | 98 | 99 |
| 86 | 91 | 96 | 99 |
| 85 | 88 | 93 | 98 |
| 84 | 85 | 87 | 97 |
| 83 | 80 | 81 | 95 |
| 82 | 75 | 73 | 93 |
| 81 | 70 | 64 | 88 |
| 80 | 64 | 54 | 83 |
| 79 | 57 | 46 | 76 |
| 78 | 50 | 38 | 69 |
| 77 | 43 | 30 | 61 |
| 76 | 38 | 23 | 52 |
| 75 | 33 | 18 | 46 |
| 72 | 20 | 8 | 29 |
| 70 | 14 | 4 | 20 |
| 65 | 4 | 2 | 8 |
| 60 | 1 | 1 | 3 |
| Section | Questions / minutes | Mean | Standard deviation | Sample | Calculator |
|---|---|---|---|---|---|
| Quantitative Reasoning | 21 / 45 | 77.92 | 6.52 | 508,910 | No |
| Verbal Reasoning | 23 / 45 | 79.43 | 4.89 | 508,711 | No |
| Data Insights | 20 / 45 | 75.14 | 5.98 | 508,575 | On-screen, built in |
All figures are GMAC's (mba.com widget and exam structure pages, data 2021-2026). GMAC does not publish how the three section scores combine into the total, only that they are weighted equally, so you cannot add your sections to predict a total, and this page does not try.
The 10th Edition (200 to 800) was sunset on 31 January 2024. GMAC says the two scales "cannot be compared" directly and gives its own warning example: 600 and 605 look similar but represent very different performance. The concordance links them only through shared percentiles. On the Aug 2026 table a 10th Edition 700 sits in the same percentile band as Focus 645 to 655, and a 750 with 695 to 715.
| 10th Edition total | Focus total(s) in the same band (official) | Percentile(s), Aug 2026 |
|---|---|---|
| 800 | 805 | 100.0 |
| 790 | 805, 795, 785 | 100.0, 100.0, 100.0 |
| 780 | 785, 775, 765, 755 | 100.0, 100.0, 99.9, 99.8 |
| 770 | 755, 745, 735 | 99.8, 99.6, 99.4 |
| 760 | 735, 725, 715 | 99.4, 98.9, 98.6 |
| 750 | 715, 705, 695 | 98.6, 97.5, 96.9 |
| 740 | 695, 685 | 96.9, 95.1 |
| 730 | 685, 675 | 95.1, 94.0 |
| 720 | 675, 665 | 94.0, 91.2 |
| 710 | 665, 655 | 91.2, 89.6 |
| 700 | 655, 645 | 89.6, 85.6 |
| 690 | 645, 635 | 85.6, 80.9 |
| 680 | 635, 625, 615 | 80.9, 78.2, 75.4 |
| 670 | 615 | 75.4 |
| 660 | 615 | 75.4 |
| 650 | 615, 605, 595 | 75.4, 69.4, 66.3 |
| 640 | 595, 585 | 66.3, 59.9 |
| 630 | 585 | 59.9 |
| 620 | 585, 575 | 59.9, 56.7 |
| 610 | 575, 565 | 56.7, 50.4 |
| 600 | 565, 555 | 50.4, 47.4 |
Schools deal with this in different ways. ISB and IIM Ahmedabad publish separate Classic and Focus figures for the same class. IIM Calcutta states its floor as a percentile and says it applies "uniformly regardless of whether the candidate's GMAT score is from the Classic or Focus Edition". A 10th Edition score stays valid for five years from its test date; since the last 10th Edition sittings were in January 2024, the latest of them run into January 2029. They can be sent while valid but are excluded from the new Superscore.
Four quick readings of the current table for the programmes most readers of this page are targeting. The school figures are the schools' own; the percentile lookups are ours.
Sources: ISB class profile, IIMA PGPX brochure 2026-27, IIMC MBAEx admission policy 2027-28, GMAC concordance Aug 2026, support.mba.com on the Superscore.
The official report gives a total, three section scores and a percentile for each. It arrives in your mba.com account within five days (GMAC says it can occasionally take up to 20), and it cannot be expedited. You do not need to cancel a score you dislike: schools do not receive an attempt you do not send, with one qualification since 12 August 2026. If that unsent attempt supplies a section of your Superscore, the section score, its date and its delivery method appear on the Superscore report that travels with any score you do send. Your five free score sends must be used within 48 hours of the score becoming available; after that each Additional Score Report is US$35. Because percentiles are recomputed each year, the percentile on the score record in your mba.com account may be lower next year than today, with the same total (GMAC says the test-taker report is not downloadable or available as a PDF). When an application form asks for a percentile, follow that programme's instructions: IIM Calcutta, for example, reads percentiles "as reported on the candidate's official GMAT score report".
ProSyllabus has no GMAT practice quizzes yet, so this guide does not point you to any. GMAC's own free Official Starter Kit on mba.com includes two full-length practice exams scored the same way as the real test.
ProSyllabus is an independent study site, not part of GMAC or of any school named here. Every rule is quoted or summarised from GMAC (mba.com, gmac.com, support.mba.com) and the schools' own admission pages; they decide every individual case.
The 85.6th percentile on GMAC's concordance published in August 2026 (data July 2021 to June 2026). The mba.com widget rounds it to 86. The 87th in a GMAC article was the July 2025 table's 86.7, rounded.
The 94.0th percentile on the Aug 2026 table (94 on the widget). It is also ISB's Focus median for both the Class of 2026 and the Class of 2027.
The 97.5th percentile on the Aug 2026 table (98 on the widget). It was 98.2 in the Aug 2024 table.
GMAC recalculates percentiles every year over a rolling five-year window and updates existing score reports automatically. Your total does not change; the ranking attached to it does.
The mean total across the 2021-2026 data is 554.94, with a standard deviation of 94.24, from 509,182 test takers. A 555 is the 47.4th percentile.
No. On the 2021-2026 data an 80 is the 54th percentile in Verbal, the 64th in Quant and the 83rd in Data Insights.
The Aug 2026 concordance places a 10th Edition 700 in the same percentile band as Focus 645 to 655. GMAC says the two scales cannot be compared directly, only through percentiles.
No. GMAC says the three sections are weighted equally but does not publish the conversion, so the total cannot be rebuilt from the sections.
No. GMAC says the Superscore is on the same 205-805 scale but has no associated percentile ranking.
Not yet. GMAC's free Official Starter Kit on mba.com includes two full-length practice exams.
official = a document published by the conducting body. reported = a news or coaching site we could not check against an original. Where sources disagree this page says so rather than picking one.