How is the GMAT scored?
The GMAT reports a Total Score from 205 to 805, built from three equally weighted section scores — Quantitative Reasoning, Verbal Reasoning, and Data Insights, each scaled 60 to 90. The exam is question-adaptive, and your score is determined by three things only: the difficulty of the questions you answer correctly, the difficulty of the questions you answer incorrectly, and the number you leave unanswered. How many you get right is not the metric, and neither is where in the section your mistakes fall.
What scale is the GMAT scored on?
| Score | Range | Interval |
|---|---|---|
| Total Score | 205–805 | 10 points |
| Quantitative Reasoning | 60–90 | 1 point |
| Verbal Reasoning | 60–90 | 1 point |
| Data Insights | 60–90 | 1 point |
- Each section contributes equally to the Total Score.
- Every Total Score ends in 5 — a 705 exists, a 700 does not. This is the fastest way to tell a current GMAT score from a legacy (200–800) score, where totals ended in 0.
- Scores are valid for five years.
How does the adaptive algorithm produce a score?
The exam estimates your ability continuously rather than adding up points:
- Each section opens with a medium-difficulty question.
- Answer correctly and the next question is typically harder; answer incorrectly and it's easier.
- Your ability is estimated successively — after question 1, then questions 1–2, then 1–2–3, and so on through the section.
- The final estimate — which questions you got right and wrong and how difficult they were — becomes your section score, and the three section scores combine with equal weight into the Total.
The successive-estimate design means no single question decides the outcome: in GMAC's own words, "mistakes (or lucky guesses) are compensated for by the rest of the questions on the test."
Do the first questions count more?
No — GMAC lists "my score will be worse if I get the first or second question of a section wrong" among the misconceptions on its official scoring page. The myth survives because early answers visibly steer the difficulty path, and that steering gets mistaken for scoring weight. Position is not one of the scoring factors: the same set of right and wrong answers produces the same estimate wherever the errors fall.
GMAC's misconception list covers four more pieces of received test-prep wisdom, all false:
| Belief | Reality per GMAC |
|---|---|
| Taking longer on a question lowers your score | Time per question is not a scoring factor |
| An answer changed from right to wrong in review counts extra | Edited answers are scored like initial answers |
| Some question types are weighted more heavily | No question-type weighting exists |
| Better to leave a question blank than guess when time runs out | Blank is the worse outcome — unanswered questions are penalized |
What happens if you don't finish a section?
An explicit penalty. GMAC's scoring page states: "If you do not finish in the allotted time, your score will be penalized reflecting the number of questions unanswered." Two practical consequences:
- Always answer everything. A guess might be right, and at worst is scored as one more wrong answer; a blank is penalized as unanswered. GMAC lists "better to leave it blank than guess" among the scoring misconceptions.
- Build the endgame into your pacing. With 45 minutes for each section (21 Quant, 23 Verbal, or 20 Data Insights questions), decide in advance the minute-mark at which you stop solving and start selecting. The Question Review & Edit tool — bookmark anything, change up to three answers per section — rewards finishing with time to revisit, not leaving questions open.
What do GMAT percentiles look like?
Percentile values from GMAC's July 2025 concordance table, based on tests from July 2020 through June 2025:
| GMAT Total Score | Percentile | Legacy (200–800) equivalent |
|---|---|---|
| 735 | 99.5% | 760–770 |
| 705 | 98.0% | 750 |
| 685 | 95.8% | 730–740 |
| 655 | 90.5% | 700–710 |
| 645 | 86.7% | 690–700 |
| 605 | 70.3% | 650 |
| 565 | 50.9% | 600–610 |
The right-hand column corrects a common misreading: a 645 can look underwhelming next to remembered "700-club" numbers, but it concords to a legacy 690–700 and outperforms 86.7% of test takers. Judging a current score against legacy-scale folklore understates it by roughly 50 points.
What does the scoring design mean for how you prepare?
- Train at and above your difficulty frontier. Since the score comes from the difficulty you can sustain, practice that never leaves comfortable territory caps the estimate the algorithm can reach.
- Rehearse full timed sections, endgame included. The unanswered penalty makes the last five minutes of each section a scored skill of its own.
- Spend zero energy on question-position anxiety. Early errors change the path, not the weight — steady accuracy over 45 minutes beats a perfect opening.
Working with score mechanics like these — knowing what the algorithm rewards and what it ignores — is a core part of how the CoPilot Prep GMAT cohort structures timed practice. For the coaching behind it, see Simon Flynn's background.