What similarity rate is acceptable?
Last updated:
Many Turkish authors ask which similarity percentage is "safe". The honest answer from the organisations that build and teach these tools is that no single percentage is. This guide explains what the report shows, why a number alone cannot decide a case, and how to prepare your text.
What a similarity report shows
Crossref's documentation for Similarity Check explains that the similarity score is the share of the words in a document that match the repositories selected for the check. Exclusion settings, such as removing the reference list or quoted text, change which matches count, and the score is recalculated when an exclusion is applied.
The same guide stresses a point authors often miss: a match is a prompt for a human to look, not evidence of misconduct. Heavy overlap can simply come from a paper that cites and quotes sources correctly.
How the journal reads the report (from the journal's policy page)
Journals set their own approach, so read the plagiarism or similarity section on the journal's own policy page before you submit; this guide cannot state a journal's rule for it. What the public guidance agrees on is how a report should be read:
- Crossref's guidance sets no acceptable threshold and advises against rejecting manuscripts automatically on the basis of a score.
- Editors and reviewers are expected to judge whether each match is acceptable, as part of normal review.
- Overlap with the authors' own preprint is expected and is not a problem in itself, whereas overlap with a preprint by different authors may deserve investigation.
So a low number does not prove a paper is clean (one copied paragraph from an unindexed source may not appear), and a high number does not prove it is copied.
Text recycling and self-plagiarism
COPE uses "text recycling" (also called "self-plagiarism") for cases where parts of an author's own earlier publications reappear in a new one. COPE's guidance asks editors to judge each case individually and not to rely on a percentage alone.
- Methods sections: overlap is often hard to avoid because a standard technique can be described accurately in only so many ways. Editors may accept similar wording when the authors are open about it and cite the earlier description.
- Introduction: reused background is treated as less serious than reused hypotheses, discussion or conclusions.
- Duplicated data or results are a separate and more serious matter, handled as redundant publication.
COPE's advice for authors is short: cite your earlier publications, be open about any overlap, reword text that is not a methods description, and do not duplicate data, results or discussion. The same applies to a thesis or a congress abstract that you are turning into an article; say so in your cover letter.
Before you submit
- Write the text in your own words from your notes, not by editing someone else's paragraph. Translating a foreign-language source sentence by sentence is still copying.
- Put every quotation in quotation marks and cite it, even a short one.
- Cite your own earlier work, including a thesis, a preprint and a congress abstract, and tell the editor about them.
- If your institution gives you a similarity report, look at the individual matches, not only the total. Note which matches are the reference list, standard method wording or common phrases.
- Rewrite where a match is real overlap with another source; do not just swap synonyms to lower the number.
- Keep the report; if the editor asks about a match, a short, honest explanation helps.