CAT 2025 tied for the second-hardest paper in five years. Here's the breakdown.

The data

204 questions across 3 slots of CAT 2025, every question tagged by section, topic, and difficulty (easy / medium / hard).

---

Overall: 53% hard

Difficulty	Questions	%
Hard	108	53%
Medium	86	42%
Easy	10	5%

CAT 2025 difficulty breakdown (204 questions)

At 53% hard, CAT 2025 tied 2022 (also 53%) as the joint second-toughest paper in five years — well above 2021 (44%) and the unusually gentle 2024 (26%), and behind only 2023, which peaked at 66% hard. If your mock performance dipped in November, difficulty was a genuine factor.

---

Five-year difficulty trend

Difficulty swings sharply from year to year. CAT 2024 was the gentlest paper in the set at 26% hard; 2025 snapped back to 53% — tied with 2022, and behind only 2023's 66% peak.

Year	Overall hard %	DILR hard %
2021	44%	88%
2022	53%	62%
2023	66%	87%
2024	26%	38%
2025	53%	76%

% hard by year — overall vs DILR

DILR is the most volatile section — it has topped 85% hard twice (2021 and 2023) and hit 76% in 2025, but cratered to 38% in the unusually easy 2024 paper. The takeaway for mock-takers: one paper's difficulty tells you little; your trend across several mocks is the real signal.

---

Section breakdown

Section	Questions	Easy	Medium	Hard	Hard %
VARC	72	3 (4%)	54 (75%)	15 (21%)	21%
DILR	66	5 (8%)	11 (17%)	50 (76%)	76%
QA	66	2 (3%)	21 (32%)	43 (65%)	65%

Difficulty distribution by section (2025, %)

The gap between sections is striking. DILR at 76% hard is a fundamentally different experience from VARC at 21% hard. Candidates who prepared each section uniformly may have hit an unexpected wall in DILR.

---

VARC: 72 questions — the gentlest section

VARC held to form. RC remained 67% of the section (48 of 72 questions), with difficulty concentrated in the medium band (73% of RC medium, 21% hard).

Topic	Questions	Difficulty skew
Reading Comprehension	48	73% medium · 21% hard
Para Jumble	6	mostly medium (33% hard)
Para Summary	6	all medium
Sentence Insertion	6	mostly medium (17% hard)
Odd Sentence Out	6	mostly medium (33% hard)

Passage topics in 2025 leaned towards humanities and social science — genres that produce more inference and interpretation questions, where the challenge is logical reasoning from the text rather than vocabulary.

---

DILR: 66 questions — 76% hard

DILR was the paper's most demanding section. Three of its four sub-types were rated 100% hard; only Data Interpretation offered any non-hard questions.

Topic	Questions	Hard %
Data Interpretation	40	60%
Arrangements	17	100%
Scheduling	5	100%
Logical Reasoning	4	100%

DILR questions by topic (2025)

Every Arrangement, Scheduling and Logical Reasoning question was rated hard — zero medium or easy across all three. In most prior years, Arrangements included a couple of entry-level questions; in 2025 that foothold was absent entirely. DI at 60% hard was also elevated — the only non-hard DILR questions sat in basic tables and bar-graph sets, while every matrix, circular and scatter-plot set was hard.

---

QA: 66 questions — 65% hard

Arithmetic and Algebra tied at 23 questions each — together they are 70% of QA.

Topic	Questions	Hard %
Arithmetic	23	61%
Algebra	23	70%
Geometry	9	89%
Number System	7	71%
Modern Math	4	0%

QA questions by topic (2025)

Geometry (89%), Number System (71%) and Algebra (70%) were the steepest topics; Arithmetic, the highest-volume topic, was comparatively the most approachable at 61% hard. Modern Math was the section's one easy pocket — 4 questions, none rated hard.

---

Three things that stand out

1. DILR's footholds were scarce.

Standard strategy involves scanning DILR sets and picking up the 1–2 easiest first. With Arrangements, Scheduling and Logical Reasoning all 100% hard, the only non-hard questions sat in basic Data Interpretation (tables and bar graphs). Scoring in DILR required sustained performance on genuinely hard sets, not cherry-picking easy ones.

2. VARC was where composure, not skill, separated candidates.

At 21% hard, VARC difficulty was not the bottleneck. Dense inference questions on unfamiliar topics under a 40-minute clock were. High scorers likely read fast and selected decisively; slower readers ran out of time regardless of ability.

3. QA's one easy pocket was Modern Math.

Modern Math ran 4 questions, none rated hard — the section's only source of quick, low-risk marks. Every other QA topic ran at least 61% hard, with Algebra (70%), Number System (71%) and Geometry (89%) the steepest. Candidates who skipped Modern Math in prep gave up the easiest points on offer.

---

What this means for 2026 prep

The clearest signal from 2025 is the DILR difficulty ceiling: a section that ran 76% hard, with non-hard questions confined to basic DI, rewards preparation on complete hard sets under full time pressure — not sampling hard questions in an untimed environment.

---

How we tag difficulty

Every question in this analysis is hand-tagged easy / medium / hard on a fixed rubric:

• Easy — direct application of a single idea, no traps.
• Medium — moderate reasoning or a 2-step calculation.
• Hard — multi-step, requires elimination, or carries a trap.

The same rubric is applied across all five years (2021–2025), so the year-on-year comparisons above are apples-to-apples. One honest caveat: this is expert editorial judgment, not candidate response data — the actual exam never publishes per-question difficulty — so read these figures as a careful, consistent estimate, not a psychometric measurement.