How to Do Legal Research in India

Most people research a legal question backwards. They hear the problem, form an instinct about what the law probably is, and go straight to a database to look for cases that confirm it. Some hours later they have thirty judgments, no clarity, and a nagging sense of having missed something. They usually have.

The method below runs in the opposite order. It is not clever and it is not fast on the first pass. It is simply the order in which the material actually depends on itself: you cannot know which law applies until you know the facts, you cannot use a case until you have read the section it interprets, and you cannot rely on any of it until you have checked it still stands.

1. Settle the facts before you open any law

Research done before the facts are fixed is research done twice. The same sentence (they have not paid me) could be a suit for recovery, a summary suit, a complaint under section 138 of the Negotiable Instruments Act, an insolvency petition, an arbitration or a consumer complaint. Each has a different forum, a different limitation period and an entirely different body of case law. Nothing tells you which until you know whether there was a written contract, whether a cheque was issued, when it bounced, and whether the debtor is a company.

So before anything else, write out what happened, in date order, with the document that proves each entry beside it. Three columns: date, event, proof.

This one table does more work than any search. It shows you when the right was breached, which is when limitation started running. It shows you the gaps: a six-month silence in the middle of a chronology is either innocent or the weakest point in the case, and you want to find it before the other side does. And any row with nothing in the third column is a fact you will have to prove by oral evidence, or abandon.

2. Turn the grievance into a legal question

A court cannot act on a grievance. It acts on a set of facts which, if proved, entitle someone to a remedy. Your job is to convert one into the other, and the way to do it is to break the claim into its ingredients and set your facts against each one.

For a breach of contract, roughly: a valid contract existed; the plaintiff performed or was ready and willing to perform; the defendant breached; loss followed from the breach. Write those down the left, your facts down the middle, your proof down the right. Wherever an element has no fact against it, you have either a gap in instructions or no case, and either way you now know what to do next, which no amount of reading judgments would have told you.

Find the ingredients in this order: the section itself, which usually sets them out on its face; then a leading judgment, because courts often summarise them in numbered form; then a standard commentary.

3. Read the statute properly, and read all of it

Most research questions are answered in the bare Act by people patient enough to read it. Judgments are what you reach for when the statute runs out, not before.

• Read the whole section, not the sentence you were sent. Provisos, explanations and exceptions frequently reverse the main clause.
• Check the definitions. Statutes define ordinary words in extraordinary ways. Substitute the definition and re-read. A very large share of statutory argument is really argument about a definition clause.
• Watch the operative verbs. “Shall” is ordinarily mandatory, “may” ordinarily enabling, but whether a requirement is mandatory or merely directory is itself a classic litigated question.
• Watch for “notwithstanding” and “subject to”. A non obstante clause gives a provision overriding effect. These few words decide conflicts between statutes.
• Confirm the provision is in force, in that form. Amendments are frequent and commencement is sometimes partial.

That last point is doing real work at the moment. The Indian Penal Code, the Code of Criminal Procedure and the Indian Evidence Act were replaced on 1 July 2024 by the Bharatiya Nyaya Sanhita, the Bharatiya Nagarik Suraksha Sanhita and the Bharatiya Sakshya Adhiniyam. Offences committed before that date are generally still dealt with under the old law, so for some years you must work with both sets of provisions and establish which applies to your facts. A textbook printed in 2022 will not tell you this.

4. Then search for case law, and search properly

Do not search your facts. Search the proposition you need established. “My client's landlord locked the shop” is a fact. “Whether a person in settled possession can be dispossessed otherwise than by due process of law” is a proposition, and only the second can be searched. Write it out as a sentence, in the language a judgment would use.

Then:

• Search by Act and section first. Far more precise than keywords, and it eliminates most of the noise.
• Use terms of art in quotation marks: “readiness and willingness”, “settled possession”, “substantial question of law”. Judgments use settled vocabulary; exploit it.
• Expect the first search to fail. Its job is to teach you the vocabulary that makes the second search work. Two or three iterations before the results sharpen is normal, not incompetence.
• Find the leading case (the one everything else cites), then move backwards to what it relied on and forwards to what has cited it since. Forwards matters more: that is how you learn whether it survived.
• Search for the answer against you. Better you find it now than in court.

You are finished when new searches return only cases you have already seen, and you can state the position in three sentences with an authority for each. Three well-chosen authorities beat fifteen. Long citation lists signal that the advocate has not decided which case actually helps.

5. Read the judgment, not the headnote

The headnote is written by the publisher's editors, not by the Court. It carries no authority whatsoever. It is a finding aid, and treating it as a substitute for reading is the most common defect in Indian legal research.

Read a long judgment in this order, which is faster and produces better comprehension than starting at the beginning: the cause title block, to see the court, the Bench strength and the date; then the last few paragraphs, to learn the outcome; then the paragraph framing the questions for determination; then the facts; and only then the reasoning, which now makes sense because you know where it is going.

What you are looking for is the ratio decidendi: the principle applied to the material facts, on which the decision rests. That binds. Everything else is obiter: persuasive at most. The working test is to remove the proposition and ask whether the outcome would change. If yes, it is ratio. If the court would have reached the same result anyway, it is obiter, and you must describe it as an observation rather than a holding.

Two further points that decide arguments. A judgment is an authority for what it decides, not for everything that logically follows from it. And judicial language is not statutory language; arguments built on the grammar of a single judicial sentence tend to fail.

6. Check it is still good law

This is a separate step from finding the case, and it is the one people skip. The most dangerous authority in any file is a strong judgment on your facts that was overruled three years ago.

• Confirm you have the right case and the correct citation. Party names recur, and there are often several judgments between the same parties at different stages.
• Check the subsequent history. A High Court decision may have been stayed, set aside or affirmed.
• Use a citator to see how it has been treated: followed, approved, distinguished, doubted, referred to a larger Bench, overruled. This is what paid databases are for. On free databases, approximate it by searching the case name and reading the recent judgments that cite it.
• Check the Bench strength against any conflicting decision. A larger Bench prevails.
• Check whether the statute it interprets has since been amended or replaced.

Citing a decision that has been overruled, or citing it without disclosing that it has been, damages your credibility for the remainder of the hearing, and the court will thereafter distrust everything else you say.

Where to look

Free and authoritative. India Code for central bare Acts with amendment history; the e-Gazette for notifications and commencement dates; the Supreme Court of India and the Digital Supreme Court Reports for judgments in their official form; eCourts for Supreme Court and High Court judgments; PRS Legislative Research for bill tracks and the policy history behind a statute; and eCourts for free full-text search across Supreme Court and High Court judgments.

Paid. The subscription databases are what most offices are built around, largely for their citators: the apparatus that shows how a judgment has been treated since it was delivered. If your chambers has a subscription you will use it. None of them is cheap, and none of them is the only way to answer the question a citator answers.

Newer tools aimed at the verification problem. Ratio, from Legal Street, covers Supreme Court judgments. You describe the point of law in ordinary English and it returns the judgments on it, ranked by how later Benches actually used each case: whether one has been overruled, partly overruled or disapproved, and which Benches declined to follow it, so you know what is coming back at you.

What matters for this chapter is how it answers. It does not reply in prose. Every result pinpoints the paragraph that decides the point and opens the judgment itself alongside it, turned to that passage and marked on the record. So there is nothing to hallucinate. You are not reading a machine's account of what the Court held, with the judgment somewhere else to be looked up later if you have time. You are reading the judgment, at the paragraph, next to the result that sent you there. The verification step that the rest of this chapter asks you to perform is simply already done, on the page, in front of you.

A caution about free databases. They are excellent for finding a judgment and poor for verifying it. They may not tell you a decision has been overruled or referred to a larger Bench, and they sometimes carry incomplete text. I have found gaps in widely used free versions of very well known judgments. Confirm against a reported or official source before you cite.

Write it down

Research that lives only in your head is not research. Produce a short note: the question in one sentence; the answer in two or three; the material facts your answer depends on; the statutory position; each authority with its citation, the proposition it establishes and why it applies; and (never omit this) what is against you and how it can be met. Finish with what you could not resolve and what further instructions you need.

Arrange the authorities by proposition, not by date. Cite paragraph numbers rather than pages: they are stable across publishers, they let the court find your point instantly, and they demonstrate that you have read the judgment.

The mistakes worth naming

Researching before the facts are settled. Relying on a headnote. Citing a judgment for a sentence rather than a holding. Ignoring Bench strength. Leaving limitation until the draft is ready. Quoting a case decided on materially different facts. Working from an out-of-date bare Act. Collecting fifteen authorities for one proposition because you have not decided which one helps.

And, increasingly: trusting a summary produced by an AI tool without opening the judgment. The Supreme Court has held that citing fake or hallucinated precedents without verification is misconduct on the part of an advocate. Use such tools to locate a judgment if you find them useful. Then read the judgment. The habit that makes this dangerous is considerably older than the technology.

The short version

Facts, in date order, with proof. Ingredients of the claim. Forum, limitation, and any pre-condition such as a statutory notice or mandatory pre-institution mediation. The statute, read whole. Then the case law, searched by section and by proposition. Then the judgment itself, read for its ratio. Then verification that it still stands. Then the note.

Nobody knows the law; there is far too much of it and it changes weekly. What separates people who are useful from people who are not is having a reliable method for finding what governs the problem, reading it accurately, and stating it honestly.

Filed under The Legal Profession