The postgraduate CLAT — the entrance exam for the one-year LLM — gets a bigger overhaul than most people expected. The headline: for the first time, you would write out an answer to a legal problem instead of only ticking boxes. This post covers the whole redesign, including the report's sample problems.
As with everything in this report, it is a proposal aimed at CLAT 2028. The 2027 PG exam is unchanged.
What the CLAT PG exam looks like today
Right now the CLAT PG is 120 multiple-choice questions in two hours, with no sections. It covers nearly every subject from the law degree: constitutional law, jurisprudence, administrative law, contract, torts, family law, criminal law, property, company law, public international law, tax, environmental law, and labour law. Wrong answers cost 0.25 marks. Each passage is drawn from a judgment, statute, or regulation, followed by questions on it.
What the committee found wrong with it
The report is blunt. Three problems stand out.
First, errors. In 2025, eight PG questions were challenged in the Delhi High Court. The Consortium withdrew five; the court then examined the remaining three, upheld one, and struck down two, ordering marks for everyone. Similar complaints had come up in 2020. This is not a one-off.
Second, rote over reasoning. A large share of questions simply tested recall. The report's examples: an extract from the Sabarimala judgment that asked which judge disagreed with a particular concurring opinion; a question asking candidates to translate the Latin maxim "lex non cogit ad impossibilia"; a question on when India's first administrative tribunal was set up; and one asking when the National Green Tribunal Act was passed. Some questions were only loosely linked to their passage at all. And in several, the answer was again sitting word-for-word in the passage.
Third, length. The passages ballooned. The report counted the total word count across the paper: about 5,404 words in 2023, roughly 8,954 in 2024 (close to a 66% jump), and still around 8,413 in 2025. At some point a passage stops testing legal reasoning and starts testing reading stamina.
The specialist LLM problem
There is a subtler unfairness too. LLMs are often specialised — in intellectual property, or insolvency, or public law and legal theory. But the current exam tests one enormous common syllabus. That quietly advantages some candidates over others. Someone preparing for a public-law track already knows constitutional, administrative, and environmental law, so the general paper plays to their strengths. Someone preparing for an intellectual property track gets tested on none of their specialism. The committee could not fully solve this — bespoke exams for each specialisation are not feasible yet — but it tried to make the single exam fairer.
The five proposed CLAT PG sections
The committee wants the PG exam split into five parts.
The first three are the same three proposed for the undergraduate exam — Logical Reasoning and Legal Aptitude, Language and Comprehension, and Quantitative Techniques. Together these would count for about half the paper, because these thinking skills matter for postgraduate study too. Then two new sections sit on top.
4. Legal Knowledge
A multiple-choice section, but a focused one. For current legal developments, the syllabus would be limited to the two calendar years before the exam — enough to keep preparation manageable and reduce the luck factor. The emphasis shifts to depth over trivia: legal principles, doctrine, and how they apply, not the names of committees or judges or the dates of cases.
The report's sample shows what this looks like. It quotes Section 27 of the Indian Contract Act (which voids agreements that restrain someone from a lawful trade), describes the tort of inducing a breach of contract, and asks whether the two are inconsistent. The answer is not a memorised fact — it is a reasoned conclusion that the two can co-exist, confirmed by cases applying the tort. Another sample gives the elements of adverse possession and asks whether a tenant who stays on after a tenancy can claim it (no, because their possession was permissive, not hostile). You reason to the answer; you do not recall it.
5. Problem Question (written)
This is the big one. For the first time, you would write a structured answer, not pick an option.
You would choose one of three supplied problems, drawn from public law, private law, commercial law, or international law. Each would be based, broadly, on an important legal development from the previous two years — though you would be free to draw on foundational principles from any time. It tests two things the report names directly: structured legal reasoning (identifying the relevant principles, applying them systematically to the facts, and telling competing arguments apart) and lucid articulation (explaining your analysis clearly, and stating both the reasons for your position and the arguments against it).
The sample problems give the flavour. One is a constitutional problem set in the fictional country of Darshania, whose law mirrors India's: Parliament passes an anonymous political-funding law using "Democracy Certificates," passes it as a money bill, and a High Court strikes it down on three grounds (voters' right to information, an arbitrary exemption from corporate donation caps, and the misuse of the money-bill route). You advise three different parties on appeal. The other is an international-law problem about four states signing a climate treaty, then making reservations, suspending obligations after an earthquake, and retaliating — and you work through whether each move breaches the treaty. These are the kinds of open, arguable problems law students actually grapple with.
How thousands of written answers would be marked fairly
This is the obvious objection, and the report has a workable answer: you do not mark everyone's.
First, only the multiple-choice sections are scored. That produces a global shortlist — roughly twice the number of available seats — of the top performers. Only shortlisted candidates' written answers get read.
Then the marking is shared out by preference. In round one, each law school marks only the written answers of shortlisted candidates who ranked it first, and makes its offers. Any school that cannot fill its seats moves to a second round, marking the answers of candidates who ranked it second (with round-one acceptances removed from the pool), and so on down the preference list until every seat is filled. Every examiner works from a common set of marking guidelines, so standards stay consistent. It is not perfect — no marking of written work ever is — but it keeps the volume manageable and the marking at least locally objective.
Why keep an entrance exam at all
The committee looked at what top universities abroad do — statements of purpose, writing samples, undergraduate grades — and decided most of it would not transfer to India. The applicant numbers are far larger (CLAT 2025 had over 16,000 PG registrations). Indian LLB grading varies so much between universities that undergraduate marks are hard to compare. And pre-written materials like statements of purpose have become unreliable now that candidates can generate them with AI. So a written exam, sat under controlled conditions, remains the fairest option — just a better-designed one.
What this means for CLAT PG aspirants
For CLAT PG 2027: no change. Keep preparing for the current MCQ paper.
For 2028 and beyond, if this goes through: start practising structured legal writing now. Laying out an argument, applying a principle to facts, and answering the counter-argument would move from a "nice to have" to a graded skill. That is worth building early, and it is exactly the kind of thinking a good LLM demands anyway.
Frequently asked questions
What is the new CLAT PG exam pattern?
The 2026 report proposes five sections: the three UG sections (Logical Reasoning and Legal Aptitude, Language and Comprehension, Quantitative Techniques) counting for about half the paper, plus a Legal Knowledge section and a written Problem Question. The exam would stay two hours. This is a proposal for CLAT 2028; CLAT PG 2027 is unchanged.
Does the new CLAT PG have a written answer?
Yes. For the first time, the proposed CLAT PG would include a Problem Question section where candidates choose one of three legal problems and write a structured answer in prose, testing legal reasoning and clear articulation rather than only ticking boxes.
What is the Legal Knowledge section in CLAT PG?
A multiple-choice section testing legal principles and their application. For current developments, the syllabus would be limited to the two calendar years before the exam, and the focus would be on depth of doctrine rather than trivia like case dates or judge names.
How would the written CLAT PG answers be marked fairly?
Only the multiple-choice sections are scored first, producing a shortlist of about twice the number of seats. Then each NLU marks only the written answers of shortlisted candidates who ranked it first, using common guidelines, with further rounds down the preference list until seats are filled.
Why is the CLAT PG being changed?
The current PG exam rewards memorising a huge syllabus, has had recurring errors and court challenges, and features very long passages. It also disadvantages students preparing for specialised LLMs. The reforms aim to test reasoning and writing that a postgraduate law degree actually needs.
Which subjects will the CLAT PG problem question cover?
The written problem would be drawn from public law, private law, commercial law, or international law, based broadly on an important legal development from the previous two years, though you can draw on foundational principles from any time.