Salami slicing is the practice of dividing the findings from a single research project into the largest possible number of separately publishable papers, typically to inflate a publication count. It sits in an uncomfortable gray zone of research publishing: segmenting a large project into multiple papers is sometimes exactly the right thing to do, but other times it is a form of redundant publication that misleads readers, inflates a CV, and can trigger a retraction. This guide sets out what actually distinguishes the two, how the Committee on Publication Ethics (COPE) and the International Committee of Medical Journal Editors (ICMJE) define and police the problem, how editors and journals actually detect it, and what authors should do at the submission stage to stay clearly on the legitimate side of the line.
What is salami slicing?
“Salami slicing” (also called salami publication, segmented publication, or the “least publishable unit” problem) refers to dividing the data or findings from a single research project into the largest possible number of separate papers, typically to maximize an author’s publication count. COPE’s guidance on the practice describes it as increasing the risk of redundancy and overlap between papers, skewing systematic reviews and meta-analyses that may unknowingly double-count the same underlying sample, and diluting the academic contribution of each individual paper. See CASRAI’s Salami Slicing dictionary entry for the operational definition and worked examples.
Salami slicing is closely related to, but not identical with, two other terms that are worth distinguishing precisely:
- Duplicate (redundant) publication — publishing work that is substantially the same as work already published elsewhere, without adequate disclosure or cross-reference. This is the more serious end of the spectrum.
- Self-plagiarism — reusing one’s own previously published text, figures, or data without attribution. Self-plagiarism is about textual/content reuse; salami slicing is about how a dataset or project is carved up across multiple outputs. The two frequently overlap in practice (a sliced paper often also recycles methods text verbatim) but they are analytically distinct problems, covered in more depth in CASRAI’s Plagiarism vs. Self-Plagiarism comparison.
Legitimate multi-paper research vs. problematic segmentation
Not every study that produces more than one paper is salami slicing. Large, multi-site trials; longitudinal cohort studies; and multi-year projects with several genuinely distinct research questions routinely and legitimately generate a series of related publications. The distinguishing questions editors and reviewers actually ask are:
- Is the research question genuinely different? A baseline descriptive paper characterizing a cohort, followed later by an intervention-outcome paper using the same recruitment infrastructure, answers two different questions and is standard practice — provided each paper cites the other and makes clear what data and participants overlap.
- Is the overlap disclosed? Legitimate companion papers cross-reference each other explicitly: the methods section says, in effect, “participants and recruitment procedures are described in detail in [citation]” rather than silently repeating or re-presenting the same sample as if it were new.
- Would a reader need to read all the papers to avoid being misled? If splitting the results across papers requires a reader to piece together multiple articles to understand what was actually found — or if any single paper, read alone, overstates what the full dataset shows — that is a strong signal of problematic segmentation rather than a legitimate multi-paper program.
- Is the primary purpose scientific or output-maximizing? Splitting one dataset into the minimum number of papers necessary to cover genuinely separate analyses is different from splitting it into the maximum number of papers possible to inflate a publication count.
A common legitimate case worth naming explicitly: the three-paper (article-based) dissertation format, where a doctoral candidate produces several distinct, publishable chapters from one program of research. This is an accepted structure in many disciplines precisely because each paper is expected to stand on a genuinely separate question, and advisor/committee oversight plus journal peer review both act as a check — it is not, by itself, salami slicing, though the same disclosure principles apply if the papers share a dataset.
COPE’s position on redundant and segmented publication
COPE addresses this problem through a dedicated position statement, two step-by-step flowcharts, and a body of real case discussions published on its website (publicationethics.org):
- Position statement: “Handling duplicated or redundant content (salami slicing)” sets out COPE’s general framework for editors.
- Flowchart: “Redundant (duplicate) publication in a submitted manuscript” — a decision tree for what an editor should do when overlap with prior work is suspected before a paper is accepted.
- Flowchart: “Redundant (duplicate) publication in a published article” — the corresponding decision tree for when overlap is discovered after publication, which routes toward correction or retraction depending on severity.
- Case discussions such as “Salami slicing/duplicate publication,” “Self-plagiarism and suspected salami publishing,” and “Dual submission, salami slicing, redundant publication, or all three?” work through real, editor-submitted scenarios and how COPE’s Forum discussed resolving them.
Across this guidance, COPE frames the distinction between salami publication and full redundant/duplicate publication as a matter of degree rather than a bright line: salami publication involves a lesser degree of overlap between papers and is often remediable through appropriate citation, cross-referencing, and disclosure to the editor of the related paper(s); full redundant (duplicate) publication involves substantial overlap in results, data, and conclusions and is treated as a more serious breach that can warrant retraction of the later paper. COPE’s own guidance material has been described, in secondary summaries of its flowcharts, as treating roughly two-thirds content overlap as a practical marker of redundant publication rather than salami publication — but editors are expected to apply judgment to the specific case rather than a mechanical percentage rule, and CASRAI has not independently verified an exact numeric threshold directly from COPE’s own primary text at time of writing, so treat that figure as indicative rather than a fixed rule. For the mechanics of how COPE’s decision trees work more generally (including its separate flowcharts on plagiarism, authorship disputes, and image manipulation), see CASRAI’s COPE Flowcharts Explained guide and the broader COPE Guidelines Explained overview.
COPE’s core recommendation to journals is procedural: have a clear, published policy on what counts as prior publication, self-plagiarism, and redundancy that would preclude publication, and require authors to declare, at submission, any related or overlapping material that is already published, in press, or under consideration elsewhere — including copies of that material where relevant.
ICMJE’s position: overlapping publications
The ICMJE Recommendations address this under “Overlapping Publications,” a section of its editorial guidance used as the baseline authorship/publication-ethics standard by most biomedical journals. Several specific positions are directly relevant to salami slicing and redundant publication:
Simultaneous submission
ICMJE states plainly that authors should not submit the same manuscript, in the same or a different language, simultaneously to more than one journal. An author who does so should expect at least prompt rejection of the submitted manuscript, and if the overlap is discovered after publication, it can be grounds for retraction. ICMJE allows two narrow, disclosed exceptions: (1) editors of two journals may pre-arrange joint or simultaneous publication when doing so serves the public interest, with disclosure to readers; and (2) independent research groups analyzing the same public dataset, or separate systematic reviews/meta-analyses drawing on overlapping evidence, are evaluated on their own merits rather than treated as duplicate submission of “the same manuscript.”
Preprints are not prior publication
ICMJE is explicit that posting a manuscript to a preprint server does not, by itself, constitute prior or duplicate publication that would block later journal submission of the same work. Its own wording states that posting a preprint “may influence a journal’s interest in or priority for peer review and publication of that work,” but does not disqualify it outright. The obligation is disclosure: authors should inform the journal that the submitted work has been posted on a preprint server, and it is the author’s responsibility — not the journal’s — to keep that preprint updated to point to the eventual published version. CASRAI covers preprint-related prior-publication questions in more depth in Preprint Server Policies: Avoiding a “Prior Publication” Conflict Before Journal Submission.
Conference abstracts and posters
ICMJE’s recommendation on overlapping publications explicitly does not prevent a journal from considering a complete report that follows an earlier preliminary presentation, such as a letter to the editor, a preprint, or an abstract or poster displayed at a scientific meeting. A conference abstract is treated as preliminary disclosure of results, not as prior publication of the full study, provided it is disclosed to the journal at submission.
Acceptable secondary publication
ICMJE also sets out conditions under which republishing an article for a different readership — most commonly a translation into another language — is legitimate rather than redundant. All of the following must be met: the authors have received approval from the editors of both journals; the priority of the primary publication is respected through a publication interval negotiated by both editors; the secondary version is intended for a genuinely different group of readers (an abbreviated version may suffice); the secondary version faithfully reflects the authors, data, and interpretation of the primary version; a footnote on the secondary version discloses that the paper has been published in whole or in part elsewhere and cites the primary reference; and the title of the secondary publication indicates that it is a secondary publication. Properly disclosed secondary publication of this kind is not redundant publication under ICMJE’s recommendations.
How journals detect salami slicing and redundant publication in practice
Detection is a mix of automated screening and editorial judgment:
- Text-similarity screening. Most journals run submitted manuscripts through similarity-detection software — iThenticate, Turnitin, or Crossref’s Similarity Check (which compares against the full-text Crossref participating-publisher corpus) — before or during peer review. High textual overlap with the same authors’ own prior publications flags for editorial review, though similarity software detects text reuse specifically and will not by itself catch salami slicing that uses freshly written text around a reused dataset. CASRAI covers how these tools actually work in Anti-Plagiarism Software: How Turnitin, iThenticate, and Crossref Similarity Check Work.
- Editor and reviewer scrutiny of methods and sample descriptions. Reviewers with domain expertise are often the first to notice that a submitted paper’s cohort, sample size, recruitment dates, or instrumentation closely match a paper they have already read or reviewed elsewhere — particularly in smaller subfields where reviewers are likely to know the relevant literature.
- Author disclosure requirements at submission. Many journals’ submission systems require a cover-letter statement confirming the work is not under consideration elsewhere and disclosing any related manuscripts, in preparation or under review, that draw on the same data, sample, or study. Failing to disclose a genuinely overlapping paper is itself treated as a breach, independent of whether the underlying segmentation was otherwise defensible.
- Post-publication literature searches and reader/reviewer tips. Overlap is also caught after publication — by readers, by researchers conducting systematic reviews who notice the same sample appearing under different paper titles, or by post-publication peer review and correspondence. CASRAI’s Post-Publication Peer Review guide covers this pathway in more detail.
How confirmed cases are handled
Once an editor confirms redundant or salami-sliced publication, the response generally follows COPE’s flowchart logic and scales to severity:
- Minor, correctable overlap (for example, an undisclosed but legitimate companion paper) is often resolved through a published correction or corrigendum that adds the missing cross-reference and disclosure, without retracting the paper.
- Substantial, undisclosed overlap that materially duplicates results, data, or conclusions is generally handled as grounds for retraction of the later-published paper, following the journal’s standard retraction process. CASRAI’s How a Retraction Actually Happens and How to Write a Retraction Statement guides cover that process end to end.
- Notification of the author’s institution. Consistent with COPE’s broader guidance on handling suspected misconduct, editors who confirm a serious case will typically notify the corresponding author’s institutional research integrity office, since deliberate redundant publication can constitute research misconduct at the institutional level, not just an editorial matter.
- Pattern findings. Where salami slicing or redundant publication appears alongside other irregularities (recycled figures, implausible turnaround times, hallmark phrasing associated with manuscript-production services), editors increasingly consider it in the context of broader integrity red flags. See CASRAI’s Paper Mills and Tortured Phrases guide and Research Misconduct Case Studies for how these patterns are investigated in aggregate.
Practical guidance for authors
- Disclose related and overlapping submissions proactively. If any other paper from the same project, dataset, or cohort is published, in press, or under review elsewhere, say so in the cover letter at submission — do not wait for an editor to discover it.
- Cross-reference companion papers explicitly in the manuscript text. The methods section of a later paper should cite the earlier paper(s) describing shared recruitment, instrumentation, or sample characteristics, rather than re-describing them as if independently derived.
- Ask whether each paper answers a genuinely separate question. If two planned papers would report substantially the same outcome measure on substantially the same sample, consider whether they should instead be combined into a single, more complete paper.
- Segmenting a large dataset is legitimate when the sub-analyses address distinct hypotheses or outcomes that would be diluted or obscured in a single omnibus paper, each paper is genuinely useful as a standalone contribution, and the relationship between the papers is transparently disclosed to editors, reviewers, and readers.
- It tends toward salami slicing when the primary driver is maximizing publication count, the papers substantially restate the same results or conclusions with only cosmetic differences, or a reader would need to consult multiple papers to avoid being misled about what the full study actually found.
- Coordinate across co-authors and collaborating sites on multi-site or multi-investigator studies specifically, since salami-slicing and duplicate-submission problems most often arise when different collaborators submit overlapping analyses to different journals without full visibility into what colleagues have already submitted.
Frequently asked questions
Is salami slicing the same as duplicate publication?
No. They sit on the same spectrum but are not identical. Salami slicing describes carving one project into the largest defensible number of separate papers; duplicate (redundant) publication describes publishing substantially the same results, data, or conclusions more than once. COPE treats salami publication as a lesser degree of overlap that is often correctable, and full redundant/duplicate publication as a more serious breach that more often warrants retraction. See CASRAI’s Salami Slicing and Duplicate Publication dictionary entries for the operational distinction.
Does publishing a preprint count as prior publication that would trigger a redundant-publication concern?
No, under ICMJE’s recommendations a preprint is not treated as prior publication. Authors must disclose the preprint at submission, and it may affect a journal’s editorial priority, but it does not by itself constitute duplicate or redundant publication.
Is it acceptable to publish a translated version of an already-published paper?
Yes, under ICMJE’s “acceptable secondary publication” conditions — both editors must approve, the secondary version must target a genuinely different readership, it must faithfully reflect the primary publication, and it must clearly disclose that it is a secondary publication with a full citation to the original.
Can splitting a dissertation into multiple journal articles be considered salami slicing?
Not inherently. The article-based (three-paper) dissertation format is an accepted structure in many disciplines, provided each paper addresses a genuinely distinct question and any shared data, sample, or methods are disclosed and cross-referenced. See CASRAI’s Three-Paper Dissertation guide.
What happens if a journal discovers salami slicing after publication?
The response scales with severity: minor undisclosed overlap is often resolved with a correction adding the missing cross-reference, while substantial undisclosed duplication of results is generally handled through retraction, following the journal’s standard process and, in serious cases, notification of the author’s institution.
