How Long Does a Retraction Take? Timeline From Notification to Correction

Researchers and institutions who have raised a concern about a published paper usually want one thing above all: a realistic sense of how long it will take. This page answers that narrowly — realistic durations, by stage, and what actually drives delay. For the substantive editorial process itself — what has to be true for a retraction to be warranted, who decides, and how a retraction differs from a correction or an expression of concern — see CASRAI’s companion guide, How a Retraction Actually Happens. This page assumes that process and focuses only on timing.

The honest answer: there is no mandated deadline, and the real-world average is measured in years, not weeks

Neither the Committee on Publication Ethics (COPE) nor the International Committee of Medical Journal Editors (ICMJE) sets a required number of days, weeks, or months for a retraction to be completed once a concern is raised — both bodies treat case complexity as too variable to fix a single number, a point CASRAI’s editorial-process guide covers directly. That absence of a mandated timeline is not the same as “fast.” Empirical studies of actual retraction dates consistently find the process runs long. One analysis of papers retracted between 2018 and 2022 found an average of 3.3 years from original publication to retraction. Other datasets have found broadly similar averages, around 2 to 2.5 years, with wide variation by field and by how the case originated — retractions tied to fast-moving, high-scrutiny areas such as COVID-19 research have been measured at well under a year on average, while retractions in some other fields have averaged well over two years.

Two figures worth distinguishing: publication-to-retraction time (how long a flawed paper sat in the literature before it was withdrawn, which includes however long it took for anyone to notice a problem) is not the same as concern-to-retraction time (how long the editorial and investigative process itself took once someone actually raised the issue). Most published studies measure the former, because it’s the figure that’s straightforward to calculate from public retraction notices — the date a concern was first raised privately to an editor is rarely public. The stage-by-stage breakdown below is about the second figure, which is what a person actively waiting on a specific case is usually asking about.

Timeline by stage: what typically drives the length of each phase

The steps themselves are described in full in How a Retraction Actually Happens; this section focuses on realistic duration and the reason each stage takes as long as it does.

• A concern is raised — effectively instant, but detection itself can take years. The point at which a reader, co-author, institution, or independent commentary platform such as PubPeer flags a problem is a single moment, not a duration — but it can occur anywhere from days after publication to many years later. CASRAI’s guide to searching the Retraction Watch Database covers how concerns and existing retractions are tracked once public.
• Editorial triage — typically days to a few weeks. The editor’s initial job is only to judge whether a concern is credible enough to pursue further, and whether it looks like a bounded error (pointing toward a correction) or something more serious (pointing toward investigation). Straightforward, author-reported errors with clear supporting evidence tend to move through this step quickly; concerns involving disputed evidence, anonymous allegations, or a reluctant author routinely take longer.
• Institutional investigation — the single biggest variable, commonly many months to several years. Where a concern touches possible misconduct rather than a simple honest error, COPE guidance directs editors to involve the author’s institution, since institutions hold the original data, lab records, and authority to interview those involved that an editor typically cannot access independently. This is the stage most responsible for the multi-year averages cited above: institutional processes have historically been able to run for years with no external deadline forcing a conclusion. COPE’s current (2025) retraction guidelines explicitly push back on this, stating that a retraction’s timeliness should not be held hostage to a stalled institutional investigation or an uncooperative author — but that is a standard for editors to apply, not a hard cap with a specific number attached, and it does not retroactively speed up an investigation already in progress.
• Interim expression of concern, if used — open-ended by design, and can itself run for years. Where credible doubt exists but the investigation hasn’t concluded, a journal can publish an expression of concern as a public marker. It is meant to be a temporary state that resolves into a correction or a retraction once the underlying investigation concludes — but because it is explicitly tied to that investigation’s own pace, an expression of concern can and does sit unresolved for extended periods in practice, which is part of why some readers encounter a paper flagged this way for years without further update.
• Determination, notice drafting, and author notification — typically weeks once the underlying investigation has actually concluded. Once an editor has enough to reach a conclusion, drafting a retraction notice that meets ICMJE’s content requirements and giving authors the customary opportunity to respond before publication is comparatively fast relative to the stages before it — though it can still extend by weeks to a few months if an author disputes the finding or requests to co-sign changes to the notice’s wording.
• Publication and indexing — days. Once a notice is finalized, registering it with a DOI and linking it to the original article through Crossref’s Crossmark update-to metadata is a mechanical, fast step. CASRAI’s guide to post-publication peer review covers that registration layer, including the Retraction Watch Database integration Crossref has run since September 2023.

Put together, the realistic shape of a contested case is: a short triage period, followed by a long and highly variable investigation phase that dominates the total, followed by a comparatively short notice-and-publication phase once a determination is actually reached. Cases resolved on the low end of the range above are typically ones where the author self-reports a clear, undisputed error and no institutional investigation is required at all — those can move from concern to published correction or retraction in weeks. Cases on the high end are almost always ones where an institutional misconduct investigation, disputed evidence, or a coordinated multi-paper case (such as a paper-mill batch) is involved.

What most often causes a case to run long

• An uncooperative or unreachable author. ICMJE guidance prefers author-initiated retraction; when an author disputes the finding, stops responding, or cannot be located, the editor can still proceed, but reaching that point takes longer than a straightforward case.
• Institutional investigations with legal exposure. Universities and research institutions generally run their own misconduct-inquiry procedures with their own timelines, often shaped by employment law, due-process requirements, and, in some jurisdictions, the involvement of legal counsel — none of which a journal editor controls.
• Disputes over data or authorship. Co-author disagreements about who is responsible, or disputes over access to underlying data needed to evaluate the claim, add rounds of correspondence an uncontested case doesn’t need.
• Coordinated, multi-paper cases. Paper-mill-linked retractions and other cases spanning many articles at once — see CASRAI’s guide on paper mills and research-integrity red flags — require batch review across dozens or hundreds of papers, and COPE’s current guidance calls for giving uninvolved co-authors in these cases a chance at re-review, which adds time by design rather than by delay.
• Journal or publisher capacity. Not every journal has a dedicated research-integrity team or a written retraction policy; publication-ethics commentary has repeatedly pointed to that structural gap — not any particular case’s substance — as a real driver of delay at smaller or less-resourced journals.

What this means in practice

If you raised a concern about a paper, or an author is waiting on the outcome of one, the realistic expectation set by the data above is months at the very fastest for an uncontested, self-reported error, and commonly one to several years for anything that requires an institutional investigation. Watching the Retraction Watch database or blog, or checking whether the journal has published an expression of concern as an interim signal, are the two most reliable ways to track a specific case’s status from outside the editorial office. For institutions running the investigation itself, CASRAI’s guide to how a research misconduct investigation actually works covers that process in more depth.

Frequently asked questions

Is there a maximum time limit for issuing a retraction after a concern is confirmed?

No. Neither COPE’s nor ICMJE’s published guidance sets a specific deadline. COPE’s 2025 guidelines discourage open-ended delay tied to a stalled institutional investigation, but that is a standard for editors to apply case by case, not a numeric cap.

Why do retractions so often take years?

Because the dominant variable is usually an institutional misconduct investigation, not the journal’s own editorial steps. Institutions control the original data and records an editor typically cannot access independently, and their processes have historically run without an external deadline forcing a conclusion — which is exactly what multi-year publication-to-retraction averages reflect.

Does an expression of concern have to resolve within a set period?

No. An expression of concern is meant to be temporary and resolve into a correction or retraction once the underlying investigation concludes, but because it’s tied to that investigation’s pace, it can remain unresolved for an extended period in practice.

Is a self-reported honest error faster to retract than a misconduct case?

Generally yes, when there is no dispute about the facts. A case where the authors themselves identify and document a clear, serious error, with no institutional investigation required, can move from concern to a published correction or retraction in a matter of weeks — far faster than the multi-year averages that reflect misconduct cases requiring institutional investigation.