A retraction statement is the notice a journal publishes to formally withdraw an article, stating under COPE’s Retraction Guidelines what was wrong, who initiated the retraction, and whether misconduct is alleged. It can be written by a journal editor, a publisher, or the authors themselves, depending on who identifies the problem.
What a retraction statement is
A retraction statement (sometimes called a retraction notice) is the formal published document that withdraws a journal article from the scientific record. It is a distinct document from the article itself: the original paper is not deleted, it is marked as retracted and remains accessible, with the retraction notice linked to it in both directions. This guide covers how to draft the statement’s content and structure. For the conceptual distinction between retracting a paper and redacting content within one, see Retraction vs. Redaction; for the broader COPE framework a retraction statement sits inside, see COPE Guidelines Explained; and for the editorial workflow that leads up to publishing one, see How a Retraction Actually Happens.
A retraction is not the only way a published record gets a correction. Journals also use corrections (corrigenda) for errors that don’t undermine the paper’s validity, and expressions of concern when an issue has been flagged but not yet resolved — for example, an institutional investigation is still underway. Writing a retraction statement for what is actually a correction-level issue, or before an investigation has concluded, is one of the most common drafting mistakes; see the mistakes section below.
Who writes and publishes it
The retraction statement is published under the authority of the journal — typically drafted or approved by the editor-in-chief, in consultation with the publisher’s legal/ethics function on higher-risk cases — regardless of whether the retraction was initiated by the authors, the editor, or the publisher. Authors do not unilaterally retract their own paper by writing their own notice; they request retraction, and the journal issues it, though the notice will typically make clear whether the authors agree with and co-signed the request.
- Author-initiated: one or more authors identify an error (e.g., a flawed dataset, a contaminated reagent, a statistical mistake that invalidates the conclusions) and request retraction. The journal still issues the formal notice, but it commonly states that the request came from the author(s).
- Editor-initiated: the editor becomes aware of a problem — often through a reader complaint, a post-publication peer review platform, image-integrity screening, or a paper-mill detection signal — and pursues retraction independently, sometimes without full author agreement.
- Publisher-initiated: at scale, particularly for coordinated paper-mill or fabricated-peer-review cases spanning many articles or journals within a publisher’s portfolio, the publisher’s integrity team may drive the process, working with (or sometimes overriding) the individual journal’s editor.
Where the retraction was contested or authors refused to agree, COPE’s guidance holds that the journal should still proceed and note in the statement that not all authors agreed to the retraction — silence from a non-cooperating author is not grounds to withhold a retraction the evidence otherwise supports.
Required elements, per COPE’s Retraction Guidelines
COPE’s Retraction Guidelines (version 2, published November 2019) set out what a retraction notice needs to contain. A compliant statement should:
- Clearly identify itself as a retraction in its heading, distinct from a correction, erratum, or expression of concern.
- Identify the article being retracted by full title, author list, and full bibliographic citation (journal, volume, issue, page/article number, DOI).
- State who is retracting the article — the author(s), the editor, or the publisher — and, where the authors did not all agree, say so explicitly.
- State the reason(s) for retraction in enough detail to distinguish honest error from research misconduct. A vague statement (“this article has been retracted at the request of the authors”) without a reason is widely criticized in the scholarly-integrity literature and falls short of current guidance, even though it remains common in older notices.
- Be linked bidirectionally to the original article — readers arriving at either the article or the notice should be able to find the other, and the retracted article should be unmistakably watermarked or labeled “RETRACTED” in every online version and format the publisher distributes.
- Be published promptly once the basis for retraction is established, and be freely accessible to all readers, not paywalled.
- Avoid other extraneous content. A retraction notice is not the place to relitigate a dispute, assign blame beyond what’s necessary to state who requested/agreed to it, or include material not relevant to explaining why the record is being corrected.
COPE also updated this guidance in 2025 to address paper-mill-scale fraud more directly — covering how to credit third-party research-integrity sleuths who identify problems (with their permission), how to handle batch or coordinated retractions with re-review for co-authors who may be innocent of any wrongdoing, and tighter expectations that a journal should not delay a retraction indefinitely simply because an institutional investigation is still pending, if the journal’s own evidence is already sufficient.
Wording: misconduct vs. honest error
The single most consequential drafting decision in a retraction statement is how it characterizes the cause. COPE’s guidance is explicit that the reason given should let a reader distinguish:
- Honest error — a genuine mistake (e.g., a mislabeled sample, a coding error in the analysis script, an unintentional duplication of a figure) that the authors could not reasonably have caught, and that invalidates the paper’s conclusions despite good-faith conduct.
- Misconduct — fabrication, falsification, plagiarism, image manipulation, undisclosed duplicate/redundant publication, or other conduct that violates research integrity norms, as opposed to a mistake.
Statements should describe what happened factually (e.g., “an image in Figure 2 was found to have been duplicated from Figure 4 of a prior publication by the same authors without attribution or acknowledgment”) rather than opining conclusively on intent unless an institutional investigation has actually made that finding. Journals routinely use hedged, evidence-anchored language — reporting what was found, not asserting a state of mind that hasn’t been formally established — while still being specific enough that the notice isn’t vague. If only some named authors are responsible for the issue prompting retraction, COPE’s guidance is that the notice should say so rather than implicate every co-author equally.
Common drafting mistakes
- No stated reason, or a reason too vague to be useful — “at the request of the authors” alone tells a reader nothing about whether the underlying findings can still be trusted.
- Conflating retraction with a lesser correction. If the paper’s central conclusions still hold after fixing the error, a correction/corrigendum is usually the appropriate instrument, not a retraction — retracting a paper for an issue that doesn’t undermine its validity is itself considered a guideline breach.
- Publishing before the process is finished. If an investigation is genuinely still underway and the journal doesn’t yet have sufficient grounds, an expression of concern — not a full retraction — is the correct interim instrument.
- Missing bidirectional linkage so a reader who finds the original PDF (especially via a mirror, aggregator, or cached copy) never encounters the retraction notice at all.
- Assigning blame more broadly than the evidence supports, implicating co-authors who had no role in and no reasonable way to detect the problem.
- Failing to update the record everywhere the article appears — abstracting/indexing databases, repository copies, and any publisher mirror should all reflect the retraction, not just the version of record on the journal’s own site.
Worked example
Retraction: “Enhanced Catalytic Yield of Compound X Under Modified Reaction Conditions” (Journal of Applied Chemistry, 2023, 41(6), 1203–1211, DOI: 10.xxxx/jac.2023.41.1203)
This article has been retracted by agreement between the journal’s Editor-in-Chief and the corresponding author, on behalf of all authors.
After publication, a reader raised a concern that the yield data reported in Table 2 could not be reproduced using the reaction conditions described in the Methods section. At the journal’s request, the corresponding author’s institution conducted a review of the original laboratory records. The review found that the spectroscopic data underlying the yield calculations for three of the five reported reaction conditions had been misattributed to the wrong experimental run during data transcription, an error the authors have confirmed and attribute to a lab-notebook transcription mistake rather than any intentional manipulation of the results. Because this affects the paper’s central quantitative claims, the conclusions of the article can no longer be relied upon.
All authors have been informed of this retraction and agree with it. The authors apologize for any inconvenience this may have caused.
This retraction notice is linked to the original article, which remains available online marked as retracted, in accordance with COPE’s Retraction Guidelines.
Note what this example does structurally: it identifies the article precisely, states who retracted it and that all authors agreed, gives a specific factual reason (transcription error, not fabrication), explains why the reason invalidates the paper’s conclusions, and confirms bidirectional linkage — all without speculating about intent beyond what the institutional review actually found.
Frequently asked questions
Who is legally or ethically responsible for issuing a retraction statement?
The journal (via its editor, typically with publisher support) issues the formal statement, even when authors initiate the request. The journal is the party with the standing to correct the version of record it published; authors alone cannot retract a paper by, for example, posting a notice on a preprint server or their own website.
Does a retraction statement need to name every specific error?
It needs to state the reason(s) in enough detail for a reader to understand why the record is unreliable and to distinguish honest error from misconduct — it does not need to reproduce an entire investigation report, and in cases involving an ongoing legal or institutional process, some detail may reasonably be withheld until that process concludes, provided the journal doesn’t indefinitely delay the retraction itself.
Can authors refuse to agree to a retraction?
Yes, and it happens. COPE’s guidance is that a journal should still proceed with a retraction the evidence supports even without unanimous author agreement, and the notice should state that not all authors agreed.
How is a retraction statement different from an expression of concern?
An expression of concern flags an unresolved issue — evidence exists that something may be wrong, but the investigation isn’t complete or the picture isn’t yet clear enough to retract. A retraction statement is the resolution: the journal has concluded the article’s conclusions can no longer be relied upon. See Expression of concern for the fuller definition.
For the mechanics of how a journal actually moves from a raised concern to a published retraction, see How a Retraction Actually Happens: The Editorial Process. For how COPE’s Core Practices frame retraction alongside its other guidance areas, see COPE Guidelines Explained. For a plain definition of retraction as a concept and how it differs from redacting content within a published work, see Retraction and Retraction vs. Redaction.
