How to Check Whether a Paper You're Citing Has Been Retracted

Papers get retracted for many reasons: the data couldn't be reproduced, an image was duplicated, the authors found a calculation error, or an investigation found misconduct. What matters for your own work is that a retraction doesn't remove the paper from the internet. The PDF stays where it was, search engines keep returning it, and anyone who downloaded it last year may never hear that it was withdrawn.

That's how retracted work ends up in new reference lists: not through carelessness, but because nothing about the paper itself looks different from the outside. The best-known example is the 1998 Lancet paper that linked the MMR vaccine to autism. It was formally retracted in 2010, and it has been cited many times since.

The good news is that checking takes about a minute per reference once you know where to look.

Retraction is not the only kind of notice

Journals use several different tools, and they mean different things for your citation.

Retraction. The paper is withdrawn because its findings or methods can't be relied on. You shouldn't cite it as support for a claim.

Expression of concern. The journal has flagged a problem but hasn't reached a conclusion yet. Treat the paper as unreliable until the situation is resolved, and check back before you submit.

Correction (or erratum). Part of the paper was fixed, such as a table, a figure label, or an author list. The paper is usually still usable, but make sure the claim you're citing wasn't the part that changed.

Retraction also doesn't automatically mean fraud. Some are for honest mistakes the authors found themselves, which is how the system is supposed to work. The retraction notice usually states the reason, so read it rather than assuming.

Where to check

The publisher's article page. A retracted paper normally carries a visible notice at the top, and the title may be prefixed with "Retracted:". Always open the page the DOI points to rather than relying on a copy you downloaded earlier.

The Retraction Watch Database. This is the most complete public record, now maintained by Crossref and free to search. You can look up a paper by title or DOI and it lists the reason for retraction where one is known.

PubMed. For biomedical papers, PubMed labels retracted articles and links the notice to the original record.

Your reference manager. Zotero can flag retracted items in your library automatically, which is useful for catching papers you added months ago and haven't looked at since. Other managers have similar features or plugins.

Check at the end, not just at the start

The easiest mistake is checking once, while you're collecting sources. A paper can be retracted between the day you add it to your library and the day you submit. If your project runs for months, run your whole reference list through a retraction check as one of your last steps before submission, and again before you send a revision.

If you've already cited one

If a paper you used turns out to be retracted, work out how much your argument depended on it. A background citation can usually be replaced with a different source that supports the same claim. If it was central to your results, you'll need to rethink the claim itself rather than just swap the reference.

There are also good reasons to cite a retracted paper deliberately, for example when you're discussing the retraction or the research that followed from it. In that case, say so in the text and mark the reference as retracted, following your style guide's convention. Citing it without saying so is the version that causes trouble.

Also watch for preprint copies. A journal version may be retracted while a preprint of the same work is still online with no warning attached, so a retraction check should include any version you cite.

Where Research Verifier fits

Research Verifier doesn't check retraction status. Every result comes from databases of real, published papers (Semantic Scholar, OpenAlex, Europe PMC, PubMed, arXiv, or your own uploads), which means the paper exists, but existing and being currently valid are two different questions. Exports include the DOI wherever one is available, so you can paste it straight into the Retraction Watch Database before you commit to citing a result. Treat that check as the last step after you've decided a paper supports your claim.