The GRE deducts nothing for a wrong answer, on any question, in any section. That single fact - confirmed directly on ETS's own scoring pages - inverts a habit that trips up candidates coming from exams like CAT or GATE, where a wrong guess can cost real marks. On the GRE, the only way to score zero on a question is to leave it blank or get it wrong; there is no third, worse outcome for attempting.
The arithmetic is simple, and it only points one way
On a question with no penalty for a wrong answer, the expected value of any attempt with a nonzero chance of being right is greater than or equal to the expected value of leaving it blank - which is always exactly zero. There is no version of this that favours skipping. Even a genuinely blind guess on a five-option multiple-choice question has an expected value of +0.20 questions' worth of credit (a 1-in-5 chance at full credit, no downside on the other four), against a guaranteed zero for a blank.
This is a different shape of problem from CAT or GATE, where a wrong MCQ answer actively subtracts marks, so a genuinely random guess on those exams is closer to break-even or negative rather than a clear net positive. On the GRE, "should I guess" is not a real question - the honest question is only ever "how much of my remaining time is worth spending trying to improve a guess I am about to make anyway."
| Situation | Expected value of leaving it blank | Expected value of a guess |
|---|---|---|
| Multiple choice, 5 options, zero information | 0 (guaranteed) | +0.20 questions' worth of credit |
| Multiple choice, 1 option eliminated | 0 (guaranteed) | +0.25 |
| "Select all that apply" (partial credit exists on some Verbal item types) | 0 (guaranteed) | Positive for any subset with real signal |
| Numeric entry, partial working only | 0 (guaranteed) | Positive for any estimate better than random |
The one place this logic needs a caveat: running out of time
The no-penalty rule applies to every question you actually answer - but a section still has a hard time limit, and there is no partial credit for an unanswered question at the end of a section either (it scores the same as an incorrect one: nothing). This means the practical risk on the GRE is not "guessing costs me marks," it's "running out of time before reaching the last few questions costs me exactly as much as getting them wrong would have." The strategic implication is the opposite of caution: if time is running low, entering a fast guess on every remaining question - rather than spending your last minute agonising over one - is strictly better, since a guess has positive expected value and an unanswered question has none.
Why this differs from an exam like GATE
GATE runs three separate marking regimes inside one paper - MCQs are penalised for a wrong answer, MSQ and NAT are not - so the correct guessing decision changes by question type within the same sitting. The GRE is simpler by comparison: there is exactly one rule, it applies everywhere, and it always favours attempting over skipping. If you have prepared for an exam with negative marking before switching to GRE prep, the adjustment is not learning a new formula - it is unlearning the instinct to hold back on an uncertain answer.
Where this reasoning is weak
- This is about marks, not time. The expected-value argument says guessing is never worse than blank for your score - it says nothing about whether the minute you spend deliberating over that guess would have been better spent elsewhere in the section. On a section-adaptive test, your overall accuracy on the first section of each measure affects which difficulty pool the second section is drawn from (see our companion post on GRE section-adaptive scoring), so rushing every question to guess-and-move-on is not automatically optimal either - it is a genuine trade-off between coverage and accuracy, not a free win.
- This is independent, general expected-value reasoning applied to ETS's published no-penalty rule - it is not GRE-endorsed test-taking advice, and ETS does not publish a specific recommended guessing strategy.
Sources
- ETS - GRE General Test Content and Structure, scoring section: ets.org/gre/score-users/about/general-test/content-structure.
- Expected-value arithmetic in this post is original, worked from the no-negative-marking rule stated on the page above - not quoted from ETS.
Logging full mocks by section lets you see whether you are actually attempting every question before time runs out, or losing marks to blanks that a guess would have turned into a real chance at credit. Karma Yogi's GRE tracker logs your section scores and timing mock over mock.
End of essay
- Anish Guruvelli