---
title: CAT Percentile vs Marks and Questions to Attempt
url: https://www.masaischool.com/resources/cat-percentile-vs-marks-and-questions-to-attempt
description: CAT percentile is a rank, not a mark. How slot normalisation works, the official sectional cut-off policy, mock vs real percentile and DILR sets to aim for.
updated: 2026-11-17
publisher: Masai (https://www.masaischool.com)
---

# CAT Percentile vs Marks and How Many Questions to Attempt: Slot Normalisation, Sectional Cut-offs, Mock vs Real Percentile and DILR Sets

> **Short answer**
>
> A CAT percentile is your position among all test-takers, not a fixed mark. CAT runs in three sessions, so raw scores are normalised first, then converted to percentiles. No official marks-to-percentile table exists, and IIM Ahmedabad needs 95 overall and 85 in each section for General and EWS. Attempt targets are estimates.

Almost every CAT aspirant asks the same two questions: how many marks make a given percentile, and how many questions should I attempt? The honest answer to the first is that nobody can say before the exam, including the coaching sites that publish tables. The honest answer to the second depends on your accuracy, and this page shows the arithmetic.

Official material is limited. The bulletin and press release give the exam structure and dates but not the question count or marking. The IIMs' normalisation note for CAT 2024 describes the scoring logic. IIM Ahmedabad's current policy gives real sectional minimums. Everything else here is labelled as a third-party report or as my own planning estimate. Last checked 16 Nov 2026.

1. **Raw marks** - Correct and wrong MCQs, plus non-MCQ answers, in each session
2. **Normalise across sessions** - Adjusts for difficulty differences between the three forms
3. **Normalise across sections** - Uses percentile equivalence between sections
4. **Scaled scores** - Per section and total
5. **Percentile** - Your position among all test-takers
How a raw CAT score becomes a percentile, as described in the IIMs' scoring note.

## Percentile is a rank, not a mark

A percentile says what share of test-takers you scored above. A 95 percentile means roughly 95 of every 100 test-takers scored below you, whatever the marks were. The same raw marks can give a different percentile in different years, because the pool and the paper change. This is why a table of marks for a given percentile from a past year is an estimate and not a rule.

The CAT bulletin says each IIM sets its own cut-offs and weights, so even a perfect percentile calculation is only half the picture. Read [how to read the result and scorecard](https://www.masaischool.com/resources/cat-2026-result-date-and-scorecard) for what the IIMs publish.

| Question | What is official | What is not |
|---|---|---|
| How many questions and marks? | Three sections of exactly 40 minutes each, no switching, 120 minutes in all | The bulletin and press release do not state the question count or marking |
| How do raw marks become a percentile? | Scores are normalised across sessions, then across sections, and scaled scores are converted to percentiles (IIM note, CAT 2024) | The exact formula for 2026 |
| What marks equal a 99 percentile? | No official table | Every published table is an estimate |
| Which cut-offs apply? | Each IIM publishes its own policy; IIMA's PGP 2027-29 table is public | Other IIMs' floors unless they publish them |

## Slot normalisation: why your friend in another slot is not scored the same way

CAT 2026 is held in three sessions on 29 November 2026. Each session has a different paper, so some will be harder than others. The IIMs' CAT 2024 note explains the response: scores are normalised to adjust for location and scale differences between the three forms, then normalised again across sections using percentile equivalence. The scaled scores that result are converted to percentiles for shortlisting.

Three practical points follow. First, raw marks from different sessions cannot be compared, so ignore comparisons with friends in another slot. Second, a harder paper does not mean a lower percentile, since the adjustment is designed to offset difficulty. Third, because the process is statistical, the result has no single mark-to-percentile line. The 2024 note says the process is similar to what other large Indian selection tests, such as GATE, use.

I could not read an equivalent normalisation note for 2026 or 2025 on the CAT site, so the 2024 note is the best official description. Third-party sites describe further detail, but treat those as explanations, not official method.

## Marking, expected marks and how many questions to attempt

Question counts and marking for CAT 2026 are not in the bulletin or the press release. A Careers360 report on the CAT 2025 pattern, which it described as unchanged from the year before, gives 68 questions (VARC 24, DILR 22, QA 22), three marks for a correct answer, minus 1 for a wrong multiple-choice answer, and no penalty for non-MCQ questions. I found no equivalent report for 2026 that I could read, so I treat the pattern as reported, not official. Confirm on the official mock on the CAT website.

Assuming +3 and -1, the arithmetic does not depend on any source. If your accuracy on MCQs is p, each attempted MCQ is worth 3p minus (1 minus p), which is 4p minus 1 marks on average.

| Accuracy on attempted MCQs | Average marks per MCQ attempt | Reading |
|---|---|---|
| 25% | 0 | Break-even: a blind guess gains nothing |
| 50% | 1.0 | Marginal; only worth it if you can eliminate options |
| 70% | 1.8 | Worth attempting |
| 80% | 2.2 | Strongly worth attempting |
| 90% | 2.6 | Attempt everything you can reach |

So the rule that matters is to raise attempts only while accuracy holds. There is no official number of questions that gives 90 or 95 percentile. If a site gives you one, it is an estimate for a past paper. Build your own target from the last five full mocks: take the number of questions you attempted and your accuracy in each section, and look for where adding questions began to cost more than it earned.

Non-MCQ answers carry no negative marking in the reported pattern, so attempt every one for which you have a method.

## Sectional cut-offs: the official policy at IIM Ahmedabad

Yes, an IIM can reject you for one weak section. IIM Ahmedabad's PGP 2027-29 policy, dated September 2026, applies minimum cut-offs on CAT 2026 percentile before it builds the shortlist for the writing test and interview. The minimums apply to the overall percentile and to each of VARC, DILR and QA separately.

| Category (non-PwD) | Overall percentile | Each section percentile |
|---|---|---|
| General, EWS | 95 | 85 |
| NC-OBC | 90 | 80 |
| SC | 85 | 75 |
| ST | 75 | 65 |

PwD candidates have lower floors in the same policy: 85 and 75 for General and EWS, 80 and 70 for NC-OBC, 75 and 65 for SC, and 65 and 55 for ST. IIMA also says the exact cut-offs depend on the applicant pool and the distribution of CAT scores and that the Admissions Committee may change them. Meeting the floor only puts you into the pool: the shortlist uses a composite of CAT percentile and an application rating. Other IIMs set their own floors, and I found no single official table that covers all of them.

The practical rule: take your lowest section percentile in recent mocks as your risk number, not your overall. A strong overall with one weak section is the pattern these floors punish.

## Mock percentile vs real CAT percentile

No official data compares mock scores with real CAT scores. A mock percentile ranks you among one provider's users on one provider's paper, in a pool that may be more or less competitive than the real one. It is not a position among CAT test-takers. Providers also differ in how hard their papers are and in how they compute percentile.

- **Use a band, not a point.** From your last five full mocks, take the lowest, the average and the highest section and total percentiles.
- **Plan on the lower half of the band.** Real-exam pressure is a reason to be cautious, though this is my advice and not official guidance.
- **Track the trend.** Change your plan only when three mocks in a row show the same weakness.
- **Test one official thing.** Take the official mock on the CAT website to learn the interface and timing.

For deciding how much to practise, see [CAT 2027 free preparation](https://www.masaischool.com/resources/cat-2027-free-preparation).

## DILR: how many sets to aim for

The official pages do not state how many sets DILR has or how many to attempt. The fixed fact is the clock: DILR gets exactly 40 minutes and you cannot move between sections. The following is a planning estimate, not data from any exam body.

| Plan (my estimate) | Time | Note |
|---|---|---|
| Scan all sets | About 4 minutes | Pick the easiest first |
| Solve set 1 cleanly | About 12 minutes | Accuracy before speed |
| Solve set 2 cleanly | About 12 minutes | Drop it if no forced deduction after 6 to 8 minutes of real work |
| Attempt set 3 only if clearly easy | Up to 10 minutes | Skip if it is not |
| Enter and check answers | About 2 minutes | Do not leave typed answers for the last minute |

Two sets solved with high accuracy beat three sets half-solved under negative marking. Where your mocks show a different pattern, trust your own numbers over this table. For a free pre-result estimate tool on a separate site, mbaonestop.com has a [percentile predictor](https://mbaonestop.com/predictor) that takes your score estimates, which is itself an estimate.

## Frequently asked questions

### What is the difference between CAT marks and percentile?

Marks are what you scored, while percentile is your position among all test-takers. After the exam, raw scores are normalised across the three sessions and across sections, converted to scaled scores and then to percentiles. Raw marks alone are never used to rank candidates, so the same marks can give different percentiles.

### How many marks for 99 percentile in CAT?

No official table exists, and the IIMs do not publish a marks to percentile list. Third-party sites give estimates based on past papers, which differ by year, slot and difficulty. Treat any figure as an estimate, and trust only your own scorecard once the result is out.

### What is CAT slot normalisation?

CAT 2026 runs in three sessions with different papers. The IIMs' note for CAT 2024 says scores are adjusted for location and scale differences across the forms, then normalised across sections, and the scaled scores are converted to percentiles. The aim is that session difficulty does not decide your rank.

### How many questions should I attempt in CAT for 90 percentile?

There is no official number. Under +3 and -1 marking, each MCQ is worth 4p minus 1 marks, where p is your accuracy, so attempt more only while accuracy stays high. Build a target from your last five full mocks: attempts and accuracy per section.

### How many DILR sets should I attempt in CAT?

No official figure exists. A planning estimate is to solve two sets cleanly and take a third only if it is clearly easy, within the 40-minute section. Drop a set if you have found no forced deduction after 6 to 8 minutes of real work. Adjust from your own mock data.

### Do IIMs have sectional cut-offs in CAT?

Many do. IIM Ahmedabad's PGP 2027-29 policy sets 95 overall and 85 in each of VARC, DILR and QA for General and EWS non-PwD candidates, with lower floors for other categories. Other IIMs publish their own, so read each admission policy.

### Is my mock percentile the same as my CAT percentile?

No. A mock percentile ranks you among one provider's users on that provider's paper. No official data maps mock scores to real CAT percentiles. Use your last five mocks as a band and plan on the lower half, rather than treating the average as a prediction.

### Does a wrong answer cost marks in CAT?

In the pattern Careers360 reported for CAT 2025, a wrong MCQ costs 1 mark against 3 for a correct one, and type-in-the-answer questions have no penalty. The bulletin does not state marking, so confirm it on the official mock exam on the CAT website.

### Can I get a CAT percentile of 97 with one weak section?

Possibly overall, but sectional floors can still rule you out at some IIMs. IIM Ahmedabad requires 85 in each section for General and EWS candidates, even with a high overall percentile. Check each target IIM's policy for sectional minimums before relying on the total.

### Where can I see the official CAT mock exam?

The CAT 2026 press release says a mock exam to understand the test format is available on the CAT website, iimcat.ac.in. It is the only official way to see the interface and timing, and it is free. The IIMs do not publish a marks to percentile table with it.

## Sources checked

- [CAT 2026 press release: sections, timing and mock](https://cdn.digialm.com/per/g06/pub/32842/EForms/image/CAT2026/CAT_2026_Press_Release_26-07-2026.pdf)
- [CAT 2026 information bulletin](https://cdn.digialm.com/per/g06/pub/32842/EForms/image/CAT2026/CAT_2026_Information_Bulletin_26-07-26.pdf)
- [IIM note on normalisation of CAT scores 2024, hosted copy](https://static.collegedekho.com/media/uploads/2024/07/30/scoring_and_equating_process_2024.pdf)
- [IIM Ahmedabad PGP 2027-29 admission policy, cut-offs](https://www.iima.ac.in/sites/default/files/2026-09/AdmissionProcedure_PGP_1109_2026.pdf)
- [reported CAT pattern: 68 questions and +3/-1 marking](https://news.careers360.com/cat-2025-exam-pattern-unchanged-all-you-need-know-about-section-wise-questions-selection-process-marking-scheme-iim-kozhikode)
