The fastest way is to paste the paper's DOI into a free checker, such as Lucent's tool at /check, which reads Crossref's metadata and answers in seconds. If you need to scan many papers at once, PubMed's "Retracted Publication" filter and Web of Science's "Retracted Publication" document type both work well. For a single paper, the ground truth is the journal's own retraction notice. Below is each method, what it actually tells you, and what none of them can.
Every paper indexed by Crossref carries a DOI, a string like 10.1038/s41586-020-2649-2. Paste it, or its doi.org link, into Lucent's free DOI checker. The tool calls Crossref's public works API live and reports whether the DOI resolves and whether Crossref's metadata carries a retraction-related field. There is no account to create and no key to request, and the result appears in seconds.
This is a single-paper lookup, not a way to browse a whole field or author for problems. For that, use the library and database methods below.
Paste it and get a plain-words answer, sourced to Crossref, in seconds.
Run the check at /checkA retracted paper is rarely deleted. Databases and publishers instead edit the visible record to flag it, most often by changing the title, for example prefixing it with "Retracted article:" or appending "(Retracted)" after the original title. Some listings also add a red banner or a watermark on the PDF itself. If you see any of these markers on a paper you were about to cite or share, treat the headline finding as withdrawn until you have read the notice explaining why.
If you are screening a reading list, an author's back catalog, or a whole topic rather than one paper, a search-time filter is faster than checking DOIs one by one. PubMed lets you filter results by publication type, and one of those types is "Retracted Publication," which surfaces retracted records inside a normal search. Web of Science offers the equivalent: a "Retracted Publication" document type you can add as a filter to any search or citation report. Both are built for institutional, high-volume use, so they assume you already have access and are comfortable with an advanced-search filter.
For browsing retractions directly rather than filtering a search, the Retraction Watch Database is the field's standard reference: a dedicated, searchable record of retractions across journals and topics, useful when you want to explore the landscape rather than check one paper you already have.
Every database, filter, and checker, including Lucent's, ultimately derives its answer from one place: the retraction notice the journal itself publishes alongside the paper. It is usually a short, separate document linked from the paper's page, stating that the paper has been retracted and, when the journal chooses to say so, why. If a database and a journal ever disagree, the journal's own notice is the fact of record.
Every method above depends on the DOI resolving and on a record existing to check against. An unresolved or unknown DOI is not the same as a clean paper. It means the check is inconclusive: the identifier may be wrong, unregistered, or simply not yet indexed. Treating "no result" as "no problem" is the single most common mistake in this kind of screening, and it fails in the direction that matters least safely. The honest read is to fail closed: an inconclusive check is not a pass, and a paper you cannot verify should be treated with the same caution as one you have not checked at all. See how to tell if a study is trustworthy for the rest of a fast sanity check, and what a retraction actually is for the full picture.
Lucent runs this exact Crossref check on every paper before it reaches an issue: DOI resolution and a scan for retraction-related fields, done before the write-up is sent. We show what was checked rather than asserting a paper is trustworthy, and an inconclusive result is reported as inconclusive, not smoothed over.
Every two days: brand-new papers in plain language, each screened for retractions and a live DOI. Three free every issue.
Check a DOI · What is a paper retraction · How to tell if a study is trustworthy · Latest issue · Privacy