CAT exam pattern at a glance
The Common Admission Test is a computer-based test conducted by the IIMs. The figures below come from the CAT 2025 notice and reports on its conduct by IIM Kozhikode; reports on the CAT 2026 notice (IIM Indore, 29 November 2026) say the pattern is unchanged. The pattern is released with each year's notification and has changed over the years, so use this page for orientation and confirm the current details on the official CAT website.
| Parameter | CAT 2025 |
|---|---|
| Mode | Computer-based test at test centres |
| Date and sessions | 30 November 2025, in three sessions |
| Total duration | 120 minutes |
| Time per section | 40 minutes, enforced separately |
| Total questions | 68 |
| VARC | 24 questions |
| DILR | 22 questions |
| Quantitative Ability | 22 questions |
| Correct answer | +3 marks |
| Wrong MCQ answer | −1 mark |
| Non-MCQ question | No negative marking |
| Maximum marks | 204 (68 questions at 3 marks each) |
Section-wise structure
VARC: 24 questions in 40 minutes
Reading comprehension passages account for the majority of the section, with the remainder drawn from verbal ability formats such as para-jumbles, para-summary and odd-sentence-out. Verbal ability questions are often non-MCQ in format, which matters because they carry no negative marking. At 24 questions in 40 minutes, the average is a little under two minutes per question, but a passage with five questions may take ten minutes by itself.
DILR: 22 questions in 40 minutes
Questions come in sets of roughly four to six, and the section usually consists of four or five sets. This is the section in which the number of questions you attempt is least relevant: what matters is how many sets you solve properly.
Quantitative Ability: 22 questions in 40 minutes
The mix typically leans on arithmetic and algebra, with number system, geometry and modern maths making up the rest. About two minutes per question is the average budget, but easy questions should take less so that harder ones can take more.
MCQ and non-MCQ questions
CAT includes questions where you choose an option and questions where you type the answer. The two behave differently:
- MCQs have four or five options, and a wrong answer costs one mark. You can eliminate options and back-solve in many cases.
- Non-MCQs have no options and no negative marking. You must produce the exact answer, so arithmetic accuracy matters more, but you lose nothing by attempting.
The numbers of each type in each section vary from year to year and are not fixed by the notice, so mocks should include both types.
The marking scheme and what it implies
With +3 for a correct answer and −1 for a wrong MCQ, the break-even accuracy for a blind attempt at a four-option MCQ is 25%: at that rate you gain 0.75 marks and lose 0.75 marks per attempt on average. In practice you are never guessing completely blind. If you can eliminate even one option, your chance rises to one in three and the expected value turns positive: (⅓ × 3) − (⅔ × 1) = +0.33 marks per attempt. With two options eliminated it is +1 mark per attempt. The practical rule: guess only after eliminating at least one option, and always attempt non-MCQs. For a worked example of the arithmetic, see accuracy versus attempts.
Worked scoring examples
Because CAT mixes MCQ and non-MCQ questions and penalises only the former, the same number of correct answers can give different scores. The examples below are illustrative, not from real results, and use the CAT 2025 scheme.
| Scenario | Calculation | Score |
|---|---|---|
| 30 correct, 8 wrong MCQ, 0 wrong non-MCQ | 30 × 3 − 8 × 1 | 82 |
| 30 correct, 4 wrong MCQ, 4 wrong non-MCQ | 30 × 3 − 4 × 1 (non-MCQ wrong answers cost nothing) | 86 |
| 40 correct, 15 wrong MCQ | 40 × 3 − 15 × 1 | 105 |
| 25 correct, 0 wrong, rest unattempted | 25 × 3 | 75 |
The second row shows why you should always type an answer for a non-MCQ question when you have a plausible one: a wrong non-MCQ answer costs the same as leaving it blank. The third row shows how an extra 15 wrong answers cost 15 marks, which equals five correct answers: if your accuracy on the marginal attempts is below one in four, you are losing marks by attempting them.
Rules that change how you should practise
- No switching between sections. You cannot use spare time from one section in another. Practise finishing each section within its limit.
- No going back after a section ends. Questions you flagged for review must be revisited before the bell.
- Multiple sessions. Because CAT is held in several sessions, scores are normalised before percentiles are computed, so the difficulty of your session is accounted for. This is why you should not judge a paper's difficulty by raw marks.
What the pattern means for your mock tests
A good CAT mock replicates the structure: three sections, 40 minutes each, no returning to a section, and a mix of MCQ and non-MCQ questions. When you pick a mock, check that it enforces sectional timing. A mock with a single 120-minute clock lets you move time between sections and hides the weakest part of your preparation.
Use sectional tests while you are building skills, then full mocks to practise combining them. Our CAT mock test strategy explains how to sequence these. The section guides for VARC, DILR and Quantitative Aptitude go into each part in depth, and the CAT mock test hub lists available tests.
How the pattern has changed
| Period | Total questions | Duration | Notes |
|---|---|---|---|
| Up to CAT 2019 | 100 (varying splits) | 180 minutes | Three sections with sectional time limits |
| CAT 2020 | 76 | 120 minutes | Shortened paper; 40 minutes per section |
| CAT 2021 and 2022 | 66 | 120 minutes | Reduced question count in DILR and QA |
| CAT 2023 onwards | 68 | 120 minutes | 24 VARC, 22 DILR, 22 QA in CAT 2025 |
The point of the table is not the exact history but the lesson: the exam has been shortened and reshaped over the years, and each change shifted the balance between speed and selection. Preparation that depends on a particular number of questions is fragile; preparation built on reading speed, set selection and arithmetic accuracy survives any redesign.
Scoring, percentiles and cut-offs
Raw scores are converted to percentiles using the performance of all candidates, after normalisation across sessions. IIMs and other schools set their own sectional and overall percentile cut-offs for shortlisting and then weight CAT with other criteria such as academics, work experience and the interview. The cut-offs vary by institute and year, and are published by the institutes themselves. Treat any single number you see on a coaching site with caution and check the admissions page of each school you are targeting.
Sessions, normalisation and why raw marks mislead
CAT is held on one day in several sessions (three in 2025), and the papers in each session differ. To keep scores comparable, the IIMs apply a normalisation procedure before computing percentiles, so a candidate who sat a harder session is not penalised. This has two implications for you. First, do not judge your performance by raw marks compared with a friend in another session. Second, in your own mocks, difficulty varies from paper to paper for the same reason, so look at trends across several mocks, not at one result.
The exact normalisation method and the weight of sectional and overall percentiles are decided and published by the IIMs and individual institutes. The CAT notification and each institute's admissions page are the places to check.
Exam-day environment and what to practise for
The exam is taken at a test centre on a computer, with the question palette on screen, a countdown timer for the current section and an on-screen calculator. Recent exam-day guidance reported by coaching sites says candidates are provided with a pen and a scribble pad at the centre and cannot bring their own rough sheets, and that the on-screen calculator is a basic one without scientific functions. Read your admit card and the instructions in the notification for the current rules. To prepare:
- Take mocks on a screen of similar size to the centre's, not on your phone.
- Practise copying figures from the screen to the rough sheet; transcription errors are common in DILR.
- Use a plain calculator window or a basic calculator while practising QA, so you are not surprised by its limits.
- Learn the palette: which colour means answered, flagged or unseen, and how to navigate within the section.
Checklist: is this mock faithful to the CAT pattern?
| Check | Faithful mock | Warning sign |
|---|---|---|
| Sections | VARC, DILR and QA in that order | Different order or extra sections |
| Timing | 40 minutes per section, enforced | One 120-minute clock |
| Question count | About 68 in total | 100 or more questions |
| Question types | MCQ and type-in questions | MCQs only |
| Marking | +3, −1 for wrong MCQ, no penalty for type-in | Uniform negative marking on every question |
| Navigation | No returning after a section ends | Free movement between sections |
The tests listed on our CAT hub show their duration and marking on each test page. A test that does not follow the pattern is still useful for building a skill, such as reading speed or DI fluency, provided you know it is not a faithful simulation.
Disclaimer and sources: The pattern above was checked in October 2026 against the CAT 2025 notice published on the official CAT website and published reports of the CAT 2025 conduct. TestMint is independent of the IIMs. Please verify the pattern in the official notification for the cycle you are appearing in.
