A retracted article is a published scientific paper that has been officially withdrawn by the journal because its findings can no longer be trusted. The journal’s editor issues a formal retraction notice when they lose confidence in the results, whether due to honest mistakes, data manipulation, plagiarism, or ethical violations. The paper isn’t deleted, but it gets flagged with a prominent “RETRACTED” label so anyone who encounters it knows the conclusions are unreliable.
Why Articles Get Retracted
Retractions generally fall into two broad categories: error and misconduct. On the error side, the authors may have made genuine mistakes in their data analysis, sampling methods, or experimental procedures. They might have accidentally omitted critical information that changes the interpretation of the results. These retractions aren’t accusations of wrongdoing. Scientists sometimes catch their own errors and voluntarily request retraction.
Misconduct is the more serious category. This includes fabricating or falsifying data, manipulating images to misrepresent results, plagiarizing other researchers’ work, or conducting research without proper ethical approval. A growing problem involves compromised peer review, where the process meant to catch flawed work before publication is manipulated. This has become especially common in certain fields. A 2025 review of AI-related retractions found that compromised peer review was the single most frequent reason papers were pulled.
Beyond these two main categories, articles also get retracted for duplicate publication (submitting the same work to multiple journals), authorship disputes where not all listed authors consented to the paper, and undisclosed conflicts of interest serious enough to undermine the work’s credibility. In some cases, the retraction notice doesn’t state a reason at all.
How the Retraction Process Works
A retraction can be initiated by the authors themselves, other researchers who spot problems, the journal’s editor, or the institution where the research was conducted. The journal editor, however, makes the final call. According to guidelines from the Committee on Publication Ethics, the editor decides to retract when they’re confident the publication is seriously flawed or misleading.
Once the decision is made, the journal publishes a retraction notice. This notice must identify the retracted article by title and authors, link to the original paper, and be freely accessible to all readers. The original article typically remains in the journal’s archive but is watermarked with “RETRACTED” across each page. Speed matters here. Journals are expected to publish retraction notices as quickly as possible to prevent other researchers from building on bad data or, in the case of medical research, health professionals from applying flawed findings to patient care.
When a journal suspects problems but hasn’t finished investigating, it can publish an “expression of concern” as an interim step. This alerts readers that the paper’s validity is in question without making a final judgment. If the investigation clears the authors, the expression of concern is removed. If it confirms problems, a full retraction follows. For smaller issues that don’t undermine the paper’s core conclusions, a correction (sometimes called an erratum) is published instead.
Retracted Papers Still Get Cited
You might assume that once a paper is retracted, it stops influencing science. That’s not what happens. Research published in BMJ Global Health found that retracted articles continue to be cited at similar rates even after retraction, and that 94% of post-retraction citations treat the paper as valid without mentioning its retracted status. This creates a chain reaction: when a new paper cites a retracted one without noting the retraction, readers of that new paper may cite it in turn, spreading unreliable findings further from the original source.
This is more than an academic bookkeeping problem. When retracted medical studies continue circulating as valid, clinicians may base treatment decisions on fabricated or erroneous data. The most famous example is the retracted 1998 study falsely linking vaccines to autism, which fueled vaccine hesitancy for decades despite being pulled for fraud.
How to Check if a Paper Is Retracted
If you’re reading a scientific paper and want to verify its status, several tools can help. PubMed and Web of Science both let you filter by publication type, including “Retracted Publication,” and they label retracted papers in search results. The Retraction Watch Database is a dedicated, searchable repository considered the gold standard for retraction data. It often includes context about why a paper was pulled that you won’t find in the standard databases.
One catch: studies have shown that retracted publications aren’t always consistently labeled across every database. A paper clearly marked as retracted in PubMed might not carry the same label if you find it through a different search engine or on the journal’s own website. If you’re relying on a study for anything important, checking its status in more than one place is worth the extra minute.
What Retraction Means for the Authors
The consequences depend heavily on why the paper was retracted. Authors who voluntarily retract due to honest errors generally face little professional fallout and may even earn respect for scientific integrity. Retractions tied to misconduct are a different story. Research published in Nature Human Behaviour found that the more attention a retraction receives, the more likely the authors are to leave their publishing careers entirely. Prior studies have documented drops in productivity and citation counts for non-retracted work by the same authors.
High-profile misconduct cases can also trigger institutional investigations, loss of funding, and termination. The reputational damage extends beyond the individual researchers, undermining public trust in the journal, the institution, and the peer review system itself.
Retractions Are Increasing
More than 10,000 papers were retracted in 2023 alone, and the annual number has been climbing steadily. Some of this increase reflects better detection. Journals have invested in plagiarism-detection software and image-analysis tools that catch manipulation earlier. But the rise also reflects real problems, including the explosion of “paper mills,” which are businesses that produce fabricated studies for sale to researchers under pressure to publish.
Retraction rates vary significantly by country and field. Saudi Arabia, Pakistan, Russia, and China have the highest retraction rates per 10,000 published papers. Engineering, electronics, and computer science fields see disproportionately high retraction numbers. The surge in AI-related publications has brought its own wave: 46% of all AI-related retractions on record occurred in 2023, following a peak in AI paper publication the year before.
A retraction doesn’t necessarily mean the research topic itself is invalid. It means that specific paper, with its specific data and methods, failed to meet the standards required for the scientific record. Other studies on the same topic may be perfectly sound. When you encounter a retracted article, the important thing is to look at the retraction notice for the reason and then check whether independent research supports or contradicts the original findings.