How to Check If a Case Is Still Good Law (With or Without Westlaw)

To check if a case is still good law, run it through a citator and read every negative-treatment case it surfaces. With Westlaw or Lexis, that is KeyCite or Shepard's. Without a subscription, pull the citing opinions on CourtListener and in Google Scholar's How Cited tab and read the most recent citing decisions yourself. Either way, the flag or the list starts the check; reading ends it.

Most advice on this question comes from the citator vendors or assumes you already pay one of them. You do not need to. The signals are worth understanding either way, because the free workflow rebuilds the same two inputs they package: your case's own history, and what later courts did with it.

What "good law" actually means

A case stops being good law in three ways, and they show up in different places:

1. Reversed or vacated on direct appeal. The case you found lost its own appeal. This lives in the case's subsequent history, and it is the most dangerous miss because the opinion you read looks intact.
2. Overruled by a later case. A court with the power to do it, often years later in an unrelated dispute, rejected the holding. This lives in the citing cases, not in your case's own history.
3. Superseded by statute or rule. The legislature or a rules committee changed the law out from under the holding. This is the hardest to catch, because the superseding authority may never cite the case at all.

Below those three sits a gray band: cases later courts have distinguished, criticized, or limited to their facts. Those are still good law in the strict sense, but their persuasive weight is bleeding out, and a prepared opponent will say so. Stanford's law library draws the same line: checking good law means checking both the subsequent history of your case and the subsequent citations to it, because each catches failures the other cannot see.

What the KeyCite and Shepard's signals actually mean

If you have Westlaw or Lexis, the citator gives you an editorial signal per case. The signals are useful precisely to the extent you know what they do and do not promise.

Thomson Reuters' own guidance says the quiet part: the yellow flag is the most common signal, and it tells you only that negative references exist, not whether they matter to your point. LexisNexis's reference page hedges the same way: the yellow triangle means citing references may have a significant negative impact. UCLA's law library is blunter: a citator is only as good as its algorithm and its editors, and you cannot rely on the flags alone, because no signal tells you whether the negative treatment touches the holding you cite or a proposition three headnotes away.

The working rule, whatever the vendor: a flag is a routing instruction, not a verdict. The only way to learn what it means for your point is to open the citing cases and read what they did.

How do you check if a case is good law without Westlaw or Lexis?

You reconstruct the citator's two inputs, history and treatment, from sources that publish them free. This is slower than a flag and more honest than one, because it forces the reading step the flag lets you skip.

Step 1 — Pull the citing opinions on CourtListener. CourtListener, run by the nonprofit Free Law Project, hosts over nine million decisions from over 2,000 courts. Every opinion page carries two lists: Authorities, the cases your case relied on, and Cited By, the later opinions that cite it. Open Cited By, sort newest first, and read down the recent entries. A court that overrules a case names the case it is overruling, so if the overruling decision is in the database, it is on this list. Our CourtListener guide walks the full interface; the citing-opinions list is the part that substitutes for a citator.

Step 2 — Cross-check on Google Scholar. Search the case name or citation, open the opinion, and click How Cited. USC's law library documents the mechanics: Scholar marks each citing case with horizontal bars showing how deeply it discusses yours, so you can start with the decisions that engaged the holding rather than string-cited it. Two limits, per the same guide: Scholar attaches no treatment signals, and Google does not disclose what its database includes, so the absence of a citing case proves nothing.

Step 3 — Search for the death words. In either tool, search the citing results for overruled, abrogated, superseded, and no longer good law. Courts use these words with discipline. The search also catches most supersession by statute, because a later court that has noticed the change will say so when it cites your case. What it cannot catch is a statute nobody has litigated yet. If your holding construes a statute or rule, pull the current text before you file; that is the one check no citation walk performs.

Step 4 — Use the citator you already pay for through bar dues. Most state bars bundle a research platform into membership. The New York State Bar Association includes vLex Fastcase free with membership, and vLex's Cert citator flags negative treatment, with human editors behind the flags. It is the closest thing to KeyCite that costs nothing, and it sits unopened in most solo practices. Our guide to the Fastcase bar benefit covers what the member tier includes and where Cert runs out; the broader no-subscription stack is in free legal research tools.

For the deeper method, reading treatment histories headnote by headnote, see the companion guide to shepardizing a case. This post is the triage; that one is the full protocol.

The pre-filing verification workflow

Before anything with a caption leaves your desk:

1. Triage the authorities. Mark the cases your argument cannot survive without. Those get the full treatment; the string-cite support gets the fast pass.
2. Fast-pass everything. Run every cited case through a citing-opinions lookup. CourtListener's Cited By list takes about a minute per case. You are scanning for reversals, overrulings, and the death words.
3. Full-pass the load-bearing cases. For each dispositive authority, confirm the subsequent history (did this case survive its own appeal?), then read the three most recent citing decisions from courts in your jurisdiction and any citing decision that discusses it at depth. Ask one question of each: does this court still treat the holding I need as law?
4. Read, don't trust. Whatever produced your list, a flag or a Cited By tab or a colleague's memo, the negative treatment gets read by the lawyer signing the filing. Every law library guide lands on the same instruction, because the flags are only as good as the editors and algorithms behind them.
5. Document the check. A line in the file, "checked 8/1, no negative treatment on the qualified-immunity holding," costs thirty seconds and answers the question a judge asks when a citation goes bad: what inquiry did you actually make? That inquiry is what Rule 11 requires. In Mata v. Avianca, the Southern District of New York fined two attorneys and their firm $5,000 under Rule 11 for a brief built on citations no one had verified, six of which turned out not to exist at all.

Where AI drafting changes the order of operations

Mata points at the new failure mode. Good-law checking used to assume the case was real and ask whether it was still alive. AI-assisted drafting adds a prior question: does the case exist at all? A language model can produce a citation with a plausible name, a real reporter, and a confident parenthetical for an opinion no court ever issued. Existence now comes before treatment, and it has to be checked mechanically, because a fabricated citation is engineered to look right. Our guide to verifying AI-generated citations covers why generated citations fail differently from miscopied ones.

The free stack has an answer here too. CourtListener's citation lookup tool checks any citation you feed it against the database, and the Free Law Project built the API behind it as a guardrail against hallucinated citations: paste a block of text and it extracts every citation and reports which ones resolve to real opinions. The tool states its own caveat: it is limited to the citations in CourtListener's system, so a miss is not proof of fabrication, only proof the citation is not in that database. Misses need a second lookup before you accuse your own draft.

This is the layer CaseRead was built for. The free Hallucination Shield takes any pasted draft, extracts the citations, and checks each one (up to 25 per run) for existence and for whether the cited source supports the sentence it is attached to. No signup, no subscription. Inside CaseRead itself, the check runs before you ever see the citation: research answers cite only sources the system actually retrieved, across all 53 US jurisdictions, and anything the system cannot verify is flagged rather than asserted. The good-law call on a living authority stays with you. The existence check reaches your desk already run, with anything unconfirmed flagged.

Run the citing-opinions pass on your load-bearing cases, read what recent courts actually did with them, and let the Shield clear the existence question in the time it takes to paste.

Frequently asked questions

What does it mean for a case to be "good law"? A case is good law when its holding still binds or persuades courts: it has not been reversed on appeal, overruled by a later decision, or superseded by a statute or rule. A case can also be weakened without dying, when later courts distinguish, criticize, or limit it. Checking good law means confirming both the case's own subsequent history and how later courts have treated its holding.

How can I check if a case is still good law for free? Pull the case on CourtListener and open its Cited By list to read the later opinions that cite it, then run the same case through Google Scholar's How Cited tab, which shows citing cases with bars indicating how deeply each one discusses your case. Search the citing results for words like overruled, abrogated, and superseded. Also check your bar association: most state bars bundle a research platform such as vLex Fastcase into membership, and its Cert citator flags negative treatment.

Is Google Scholar reliable for checking whether a case is good law? Google Scholar shows you the citing cases but does not classify them. Unlike Shepard's or KeyCite, it attaches no editorial signals telling you whether a citing case followed, criticized, or overruled yours, and Google does not disclose which cases its database includes. It is a solid starting point and a good cross-check, but for a case your argument depends on, read the citing opinions yourself or confirm through a citator.

What do the red and yellow flags on Westlaw mean? In Westlaw's KeyCite, a red flag warns that the case is no longer good law for at least one point, typically because it was reversed or overruled. A yellow flag means the case has some negative treatment, such as criticism or limitation by later courts, but has not been reversed or overruled. A blue-striped flag means the case has been appealed to the U.S. Court of Appeals or the U.S. Supreme Court. The flags are alerts, not conclusions: you still have to read the treatment to learn whether it touches the point you cite.

What happens if you cite a case that is no longer good law? At minimum, opposing counsel or the court will surface the negative treatment and your credibility on every other citation drops. At worst, it becomes a sanctions issue: Rule 11 requires a reasonable inquiry into the law behind every filing, and in Mata v. Avianca a federal judge fined attorneys $5,000 under Rule 11 for citing cases that did not exist at all. Between those poles sits the quiet cost of building an argument on authority the court must disregard.