How to Do Legal Research in Canada: A Practical Guide

Good legal research is a repeatable process, not a talent. Lawyers who seem to find the governing case in twenty minutes are not guessing well; they are running the same five steps every time, in order, and refusing to skip any of them. This guide walks through that process for Canadian law in real depth: framing the issue, finding the governing statutes, finding and actually reading the case law, confirming it is still good law, and knowing where AI tools genuinely help at each stage and where they do not.

Step 1: Frame the legal issue precisely

Research fails at the start more often than at the end. A vague question produces vague results, so the first job is converting messy facts into one or more precise legal questions, each identifying the area of law, the jurisdiction, and the specific element actually in dispute.

Take a worked example. Your client was stopped by police in Ontario at 1:00 a.m. after allegedly drifting between lanes. The officer smelled alcohol, made a roadside breath demand, and the client eventually provided samples at the station showing a blood alcohol concentration over the legal limit. He was charged under section 320.14 of the Criminal Code. He tells you he asked to call his lawyer at the roadside and was told to wait, and that a long time passed before the samples were taken. Sixteen months later, no trial date is in sight.

"Can we beat the impaired charge" is not a researchable question. Framed properly, this file contains at least three:

1. Did the police conduct at the roadside infringe the client's right to counsel under section 10(b) of the Charter, and if so, should the breath readings be excluded under section 24(2)? The governing framework for exclusion is R. v. Grant, 2009 SCC 32.
2. Were the statutory preconditions for the breath demand and the timing of the samples under the Criminal Code's conveyance offence provisions actually met on these facts?
3. Is the accumulating delay approaching the presumptive ceiling for unreasonable delay under section 11(b), as set by R. v. Jordan, 2016 SCC 27?

Notice what framing did: each question now names a jurisdiction (Canada, prosecuted in Ontario), a source of law (the Charter, the Criminal Code), and a specific disputed element. Each one can be researched, and answered, on its own.

Step 2: Understand your sources: primary vs secondary

Canadian legal materials divide into two families, and the distinction drives how you use them.

Primary sources are the law itself: constitutional documents including the Charter, federal and provincial statutes, regulations, and the decisions of courts and tribunals. Only primary sources bind anyone. Everything you eventually cite to a court as authority should be primary.

Secondary sources explain the law: textbooks, annotated codes, law review articles, and practice commentary. They bind no one, but a good secondary source is often the fastest map of an unfamiliar area, because someone has already organized the leading cases and flagged the live controversies. The classic workflow is to start secondary to get oriented, then move to primary to build the actual argument. Skipping the orientation step is how researchers end up reading forty cases to find the three that matter.

Step 3: Know the court hierarchy and what binds whom

Not every case you find carries the same weight, and sorting authorities by weight is half the craft. The Canadian hierarchy runs, broadly: the Supreme Court of Canada at the top; then the appellate courts (each province's Court of Appeal, and the Federal Court of Appeal in the federal system); then the superior trial courts (in Ontario, the Superior Court of Justice); then the provincial and statutory courts (such as the Ontario Court of Justice) and administrative tribunals.

The rule of precedent, or stare decisis, means a court is bound by decisions of the courts above it in its own hierarchy. For your Ontario impaired-driving file: Supreme Court of Canada decisions like Grant and Jordan bind every court in the country; Ontario Court of Appeal decisions bind Ontario trial courts; and a decision of, say, the Alberta Court of Appeal is merely persuasive in Ontario. Persuasive authority still matters, especially where your own province has not spoken, but you must know which kind you are holding. A common and costly mistake is building an argument on a trial-level decision from another province while a binding appellate decision on the same point sits unread.

Step 4: Find the governing legislation

Most Canadian legal questions start with a statute. Criminal law is federal, so the Criminal Code governs offences nationwide, alongside related federal statutes. Civil matters are mostly provincial: Ontario files will pull you into statutes like the province's limitations, family law, and employment standards legislation, plus the rules of court and their associated case law. Administrative decisions bring in the enabling statute of the tribunal and the standard-of-review framework from Canada (Minister of Citizenship and Immigration) v. Vavilov, 2019 SCC 65.

Two practical habits matter here. First, always work from the current consolidated version of the statute, and check whether the provision was amended between the events in your file and today; offences are governed by the provisions in force at the relevant time. The impaired-driving example is a live illustration: the conveyance offences were substantially restructured in 2018, so older case law discusses provisions that have since been renumbered and rewritten. Second, read the definitions section. Statutory definitions quietly decide cases, and they are the first place a careful opponent will look.

Official consolidated statutes are free online through the federal and provincial legislation sites and CanLII. If you are using an AI tool, statute section lookup is one of the places it saves real time: Casescout, for example, can pull the text of a Criminal Code section and answer questions about how courts have applied it, with the cases cited.

Step 5: Find and actually read the case law

With the statutory frame in place, you need the decisions that interpret it. Searching is covered in depth in our guides to Canadian case law search and finding a case on CanLII; what matters here is what you do once you have a shortlist.

Headnotes are not the law

A headnote is an editor's summary, useful for triage and nothing more. The binding part of a decision is its ratio decidendi: the legal principle actually necessary to the result. Everything else, however quotable, is obiter, persuasive at best. The only way to find the ratio is to read the decision, or at least the analysis section, and ask: what rule did the court have to adopt to reach this outcome on these facts?

Cite properly: neutral citations and pinpoints

Since the early 2000s, Canadian courts assign neutral citations in the form R. v. Jordan, 2016 SCC 27: year, court code, sequential decision number, independent of any commercial reporter. Older cases are cited to reporters, most importantly the Supreme Court Reports, as in R. v. Oakes, [1986] 1 S.C.R. 103. When you rely on a specific passage, record a pinpoint cite to the paragraph as you read. Reconstructing pinpoints at midnight before filing is a self-inflicted wound every researcher commits exactly once.

Read enough cases, in the right order

For each framed issue, read the leading Supreme Court authority first, then your province's appellate gloss, then a handful of recent trial decisions applying the test to facts like yours. In the worked example, that means Grant itself for the section 24(2) framework, then Ontario Court of Appeal authority on right-to-counsel breaches in impaired-driving investigations, then recent trial rulings on facts resembling your client's. The trial decisions rarely change the law, but they tell you how the test actually plays out, and what evidence you will need.

Step 6: Note up: confirm it is still good law

Never cite a case you have not noted up. Noting up means checking a decision's subsequent history and treatment: Was it appealed, and did it survive? Have later courts followed it, distinguished it, criticized it, or overruled it? A case can look authoritative and be quietly dead. CanLII's citator tools list the later decisions citing your case for free; the commercial platforms layer editorial treatment flags on top. Whatever tool you use, actually open the most important citing decisions; a flag is a hint, not a conclusion.

Note up your statutes too: confirm the provision is in force, check for pending amendments, and look for constitutional challenges. A provision can be on the books and nonetheless have been read down or struck under the framework from Oakes.

Where AI fits at each step

AI research tools have changed the economics of this process, but unevenly, and it pays to know exactly where they help.

• Framing (step 1): moderately useful. An AI tool can help you see which legal issues a fact pattern raises, but the judgment about what is actually in dispute is yours.
• Orientation (steps 2 and 3): strong. Asking a plain-English question and getting a cited overview of the governing framework compresses hours of secondary-source reading into minutes, provided every claim links to a real decision you can verify.
• Finding law (steps 4 and 5): strongest. This is the retrieval problem AI is built for: turning a natural-language question into a ranked shortlist of on-point authorities and statute sections. Tools like Casescout answer only from a curated corpus of real Canadian case law and legislation, never from an LLM's memory, and its verified mode mechanically checks every quoted passage against the full judgment text. General chatbots, by contrast, are dangerous precisely here; see our comparison of the best AI legal research tools in Canada for why raw LLMs fabricate citations.
• Reading and noting up (steps 5 and 6): assistive only. AI can summarize judgments and surface citing cases, but extracting the ratio and confirming good law remain your professional responsibility. No tool signs the factum.

Common mistakes to avoid

• Researching before framing. Hours disappear into keyword searches that a sharper question would have avoided.
• Citing headnotes or AI summaries as if they were the decision. Read the analysis; find the ratio.
• Ignoring the hierarchy. A persuasive out-of-province trial decision is not an answer to binding appellate authority.
• Using an outdated statute. Check the version in force at the relevant time, especially in areas recently reformed.
• Skipping the note-up. The most embarrassing moment in litigation is learning from opposing counsel that your lead case was overruled.
• Trusting uncited AI output. If a tool cannot show you the real decision behind a claim, the claim does not exist for legal purposes.

The process is the protection: frame, orient, find, read, verify. AI makes the middle dramatically faster; the ends remain craft. If you want to see what the modern version of this workflow feels like, try Casescout free, no credit card required, and run one of your own framed questions through it.

Frequently asked questions

What is the difference between binding and persuasive authority?

Binding authority comes from a court above yours in the same hierarchy and must be followed: Supreme Court of Canada decisions bind all Canadian courts, and a provincial Court of Appeal binds the trial courts of that province. Persuasive authority, such as decisions from other provinces or lower courts, may be adopted for its reasoning but does not have to be followed.

What is a neutral citation?

A court-assigned, publisher-independent citation in the form year, court code, decision number: for example, R. v. Jordan, 2016 SCC 27 is the 27th decision of the Supreme Court of Canada in 2016. Older cases are cited to printed reporters, such as R. v. Gladue, [1999] 1 S.C.R. 688.

How do I confirm a case is still good law?

Note it up: check its appeal history and read how later courts have treated it. CanLII lists citing decisions for free, and commercial citators add editorial treatment flags. Always open the most significant citing cases yourself rather than relying on a flag alone.

Can I do serious legal research using only free tools?

Largely, yes. CanLII plus official legislation sites cover Canadian primary law well, and free AI tiers now handle much of the first-pass synthesis. What free tools lack is editorial commentary and deep secondary sources, so complex or unfamiliar areas may still justify paid resources or library access.

Is it safe to use AI for legal research?

It depends entirely on the architecture. Tools that retrieve from a verified database of real Canadian decisions and cite everything they say are safe to use as a starting point, because every claim can be checked. General-purpose chatbots that answer from memory can invent cases and misstate holdings, and several courts have already confronted filings containing fabricated citations. Whatever tool you use, verify before you file.