Negative marking and accuracy in NEET

NEET's marking looks gentle: four marks for a correct answer, one deducted for a wrong one. Compared with exams that deduct a third of the question's value, NEET's deduction is only a quarter of it. That changes the arithmetic of guessing, and many students either over-guess because "the penalty is small" or under-attempt because "negative marking is dangerous".

Both are mistakes. This guide works through the numbers so you can decide on evidence, then shows why accuracy, more than attempts, decides ranks at the top.

The rules

• Correct answer: +4
• Wrong answer: −1
• Unanswered: 0
• A question with more than one correct option: +4 to anyone who marked any correct option.
• A dropped or wrong question: +4 to everyone who appeared, whether they attempted it or not.

These are from the NEET (UG) 2026 Information Bulletin. See the NEET UG exam pattern for the rest of the scheme.

Expected value, in plain terms

"Expected value" is what you gain on average per question if you face the same situation many times.

The break-even point

Set the expected value to zero: 4p − (1 − p) = 0, so 5p = 1 and p = 0.2.

Any attempt with a better than 20% chance of being right gains marks on average. A true random guess among four options has a 25% chance, so on paper even a blind guess is slightly positive.

Why "on paper" matters

The table assumes two things that are often false in the exam hall.

First, that your eliminations are correct. If you rule out the option that was actually right, your "two left" was really zero in two. Honest elimination needs a reason you could state: a fact that contradicts the option, a unit that doesn't match, a trend that runs the wrong way.

Second, that a blind guess is random. It often isn't. NEET options are written with deliberate distractors: a common misconception, a unit error, a swapped term. When you "guess" without knowledge, you are often drawn to exactly those. Your real chance on a gut-feel guess can be below 25%, and even below the 20% break-even.

Worked example: the spread of blind guesses. Suppose a candidate blind-guesses 10 questions. On average they expect 10 × 0.25 = 2.5 marks. But the outcomes vary widely:

• 3 right, 7 wrong: 12 − 7 = +5
• 2 right, 8 wrong: 8 − 8 = 0
• 0 right, 10 wrong: −10. The chance of this is 0.75 to the power 10, about 5.6%, roughly 1 in 18.

A small average gain, a wide spread, and a real risk of losing marks. At high scores, where one mark can shift your rank by hundreds of places, that spread matters.

Accuracy decides ranks

A careless wrong answer costs 5 marks compared with a correct one: the 4 you did not earn and the 1 deducted. That is why accuracy is worth more than extra attempts at the top end.

(Illustrative figures.)

B beats A by 25 marks with only 5 more correct answers, because each of those five replaced a wrong answer: 5 × 5 = 25. C left 20 questions blank and still scored almost as much as A, who attempted everything.

Accuracy also matters in ties. Under the 2026 bulletin, if two candidates have the same total, ties are broken first by biology marks, then chemistry, then physics. If they are still level, the candidate with the lower proportion of incorrect to correct answers ranks higher.

Your own numbers matter more than any rule

Every candidate's guessing accuracy is different, so measure yours. In each full mock, mark every attempted question:

• S for sure
• T for two options left
• G for a guess

Afterwards, calculate your accuracy in each group.

Worked example. After five mocks, an aspirant finds:

Their two-option calls are right less often than 50%, but still clearly profitable. Their guesses are below the 20% break-even and are losing marks. The rule for this aspirant is clear: keep attempting T questions, stop blind guessing, and work on why "sure" answers are wrong 6% of the time.

Elimination that works in NEET

• Units and dimensions. In physics and physical chemistry, options with the wrong units or dimensions can be removed at once.
• Limiting cases. Put in an extreme value (zero mass, infinite resistance, a very long time) and see which options still make sense.
• Order of magnitude. A rough estimate often rules out two options in a numerical.
• NCERT wording. In biology, an option that changes one term of a line you know (a location, a number, an example) can be ruled out with confidence.
• Statement questions. Judge the statement you are surest of first. It often eliminates several options at once.

A practical attempt rule

1. Attempt every question you are sure of, and read it slowly enough not to lose it to a misread.
2. Attempt when you have ruled out at least one option for a real reason. The arithmetic is on your side.
3. Leave pure guesses blank unless your own mock data show your guesses beat 20%.
4. Revisit, do not re-guess. In the last minutes, go back to marked questions where you have narrowed the options, not to blanks you know nothing about.

Practice

1. What is the minimum chance of being right that makes an attempt worthwhile on average?
2. A candidate has 12 questions narrowed to two options each. What is the expected gain from attempting all of them?
3. A candidate's mock data show their guesses are right 18% of the time. What is the expected value of a guess? Should they guess?
4. Two candidates each score 600. P has 152 correct and 8 wrong; Q has 153 correct and 12 wrong. If their subject marks are also level, who ranks higher?

Answers

1. 20%. From 4p − (1 − p) = 0.
2. About 6 right and 6 wrong: 24 − 6 = +18 marks expected.
3. 0.18 × 4 − 0.82 × 1 = 0.72 − 0.82 = −0.10. On average it loses marks, so no.
4. P: 608 − 8 = 600. Q: 612 − 12 = 600. Ratio of wrong to correct: P 8/152 ≈ 0.053, Q 12/153 ≈ 0.078. P ranks higher.

What to do next

• Mark S, T and G on every attempted question in your next three mocks.
• Calculate your accuracy in each group and write your personal attempt rule.
• Practise elimination on 20 physics numericals using units and limiting cases only.
• Add every careless error to your error log, with the reason.
• Combine this with NEET time strategy so your second round goes to the right questions.