CAT Score vs Percentile: 2023 to 2025 Figures and How Scaling Works

"How many marks for 99 percentile?" is the most searched CAT question, and it has no fixed answer. The marks needed move by close to 20 from one year to the next. This page explains how your raw marks become a scaled score and a percentile, and shows what those numbers looked like in the last three CATs.

Three numbers on a CAT scorecard

• Raw score: your marks from the answer key. +3 for a correct answer, −1 for a wrong MCQ, 0 for a wrong TITA or a skipped question.
• Scaled score: your raw score adjusted for the difficulty of your session. Every section is scaled separately, and the overall scaled score is the sum of the three.
• Percentile: where your scaled score ranks among everyone who took CAT that year, across all three sessions.

How scaling works

CAT runs in three sessions with different papers, so one session can be harder than another. The IIMs correct for this with a method they published for CAT 2024 in a document titled "Scaled Score Calculation". The current CAT 2026 FAQ does not describe scaling; CAT 2025 coverage, such as Sarvgyan's, describes the same mean, standard deviation and top 0.1% approach.

For each section, separately:

1. For each session, find the mean and the standard deviation of raw scores. Add them. Call this the session's base mark.
2. Find the same base mark for all candidates together, across the three sessions.
3. For each session, find the average raw score of its top 0.1% of candidates.
4. Find the same top 0.1% average for all candidates together.
5. Stretch your raw score so that your session's base mark lines up with the overall base mark, and your session's top 0.1% lines up with the overall top 0.1%.

As a formula: scaled score = (R − G1) × (T − G) ÷ (T1 − G1) + G, where R is your raw score, G1 and T1 are your session's base mark and top 0.1% average, and G and T are the overall ones. If the formula gives more than 100 for a section, the score is cut to 100.

Here is the effect, with made-up numbers for one section. Both candidates scored 40 raw. The overall base mark is 22 and the overall top 0.1% average is 62.

*Illustrative numbers, not real CAT data.*

The same raw score is worth more in a harder session. That is the whole purpose: your slot should not decide your percentile.

How the percentile is calculated

Per the same official document, the IIMs count everyone who appeared (N) and rank them by scaled score (r). Candidates with the same score share a rank. Then:

percentile = (N − r) ÷ N × 100, rounded to two decimals.

Anything from 99.995 upward rounds to 100. Sectional percentiles are worked out the same way, using the section's scaled score.

Careers360 reported that about 2.58 lakh candidates appeared for CAT 2025. At that size, 99 percentile means roughly 2,580 candidates scored higher than you.

CAT score vs percentile, 2023 to 2025

Overall scaled scores, as reported by iQuanta:

Cracku's post-result figures are close at most levels. For the 99th percentile it reports 84.8 (2025), 95.13 (2024) and 76.15 (2023). The largest gap is at the 98th percentile in 2025, where Cracku reports 76.

Three things stand out:

1. The 99 percentile mark moved by about 19 marks between 2023 and 2024, and fell by about 10 in 2025. There is no single "CAT 99 percentile marks" figure.
2. At the top, a few marks are a big jump. In 2025, the gap from 99 to 99.5 was about 8 marks; from 95 to 99 it was about 22.
3. A scaled score near 85 out of 204 reached 99 percentile in 2025. In raw terms that is roughly 28 correct answers with no errors, out of 68, though scaling shifts it a little.

Section-wise: CAT 2025

Sectional scaled scores, as reported by Cracku:

Add up the 99 percentile row and you get about 101, well above the overall 99 percentile score of 84.8. Sectional percentiles are separate rankings. Few candidates are in the top 1% of all three sections, so you can reach 99 overall without 99 in any section. Most IIMs also set sectional cut-offs, so a very weak section can still hurt.

What a CAT percentile predictor can and cannot do

A predictor takes your raw score from the answer key and maps it to last year's table. That gives a range, not an answer, because:

• your raw score is not your scaled score, and scaling depends on your session;
• the table changes every year with the paper's difficulty, as the 2023 to 2025 figures show;
• early tables are often built from small groups of candidates who share their scores.

Use a predictor to set a band, for example 98 to 99.3, and then wait for the scorecard.

The same caution applies to mock percentiles. A mock percentile compares you with the people who took that mock, who are usually more prepared than the average CAT candidate. Track your sectional accuracy and attempts across mocks instead. Those are the numbers you control.

Measure where you stand now

Before CAT on 29 November, a full-length mock tells you more than any table. Take the CAT Full Mock 1, free and read your section-wise report. More papers are on the CAT page, and the last-30-days plan shows how to use the remaining weeks.