Turnitin0

Plagiarism Percentage for Thesis

Direct answer

There is no single universal plagiarism percentage that is "safe" for a thesis, but most universities treat an overall similarity score of roughly 10%–15% or lower as acceptable at postgraduate level, with any single source usually capped around 5%. The percentage is a measure of matched text, not a verdict on intent — so a 12% score built from correctly cited quotations can be fine, while a 9% score dominated by one uncited source can trigger an investigation. Always judge your figure against your own institution's stated policy and the match list behind the number [1].

What Is an Acceptable Plagiarism Percentage for a Thesis?

"Acceptable" is defined by your institution, not by a global standard, and thesis-level thresholds are typically stricter than those applied to undergraduate coursework. Turnitin's own guidance is explicit that the similarity score is an indicator to be interpreted, and that no single cut-off applies across disciplines, institutions, or levels of study [1]. That is why two students can submit 11% and receive completely different outcomes: the number is only meaningful next to the matches that produced it [1].

In practice, most postgraduate regulations cluster around a small number of bands. An overall score in the low single digits is almost always read as clean. Somewhere between about 10% and 15% is commonly tolerated where the matches are quotations, standard methodology phrasing, or reference-list overlap. Above roughly 20%–25%, examiners usually expect a written explanation, and above that many departments escalate to a formal academic-integrity review [2]. Treat these as orientation points, not entitlements — your programme handbook overrides any general figure you read online [2].

The distribution of matches matters more than the total. A thesis at 14% spread across forty small, correctly cited matches is a very different document from one at 9% where a single monograph contributes seven percentage points [2]. Examiners and supervisors routinely open the match list before forming a view, and a concentrated block of unacknowledged text is the pattern that causes trouble [2].

Discipline also shifts expectations. Law, history, and literature theses quote primary sources heavily and often carry a higher baseline than mathematics or engineering theses, where quoted material is rare and any sustained match is conspicuous [1]. If your field is quote-dense, ask your supervisor what figure they consider normal for your department before you start revising anything [2].

How Is the Plagiarism Percentage for a Thesis Calculated?

The similarity score is a ratio: the number of words in your thesis that match text in the comparison database, divided by the total word count of the submitted document [3]. It is not a plagiarism score, and it does not assess whether you intended to copy — it reports textual overlap and nothing more [3]. A 30,000-word thesis with 3,000 matched words returns 10%, regardless of whether those words sit inside quotation marks [3].

The comparison database determines what can possibly be matched. Turnitin checks submissions against three corpora — student papers previously submitted to its repositories, academic publications and journals, and the open web — so a thesis drawing on a heavily digitised literature base will naturally accumulate more matches than one working from archival or field data [3]. This is also why the same chapter can score differently on different days: the database grows as new papers and publications are indexed [3].

Exclusion settings change the number substantially, and you should know which settings produced the figure you are looking at. Excluding quotations, the bibliography, and small matches (commonly set to exclude matches under 1% or under a set word count) can move a thesis from the mid-twenties into single digits without a word of the text changing [3]. Institutional accounts often apply these exclusions by default, while free or third-party checkers frequently do not — so a figure from one tool is not directly comparable with a figure from another [3].

Similarity and AI detection are separate metrics that appear on the same report. The similarity score measures overlap with existing text; the AI writing indicator estimates the likelihood that text was machine-generated, and Turnitin displays it as an asterisk rather than a precise number when confidence falls below its threshold [3]. A thesis can show 8% similarity and still carry an AI flag, or 22% similarity with no AI signal at all — reading one figure as a proxy for the other is a common and costly mistake [3].

How Can You Check and Lower Your Thesis Plagiarism Percentage Before Submission?

Most inflated thesis scores come from citation mechanics rather than deliberate copying. Over-long verbatim quotations, patchwriting (keeping a source's sentence structure while swapping synonyms), and block-quoting without quotation marks account for the majority of matches students are surprised by [4]. Fixing these is legitimate editorial work: shorten quotations, paraphrase at the level of argument rather than wording, and make sure every borrowed idea carries an in-text citation in your department's style [4].

Self-matching is the other frequent culprit, and it is easy to overlook. If you re-check a revised draft through a repository-based system, your earlier submission is already in the student paper database, and the new check can match your own previous text — producing a score that reflects your own draft rather than any external source [4]. This is one reason students see a figure climb between drafts and panic unnecessarily; the match list will show the earlier submission as the source [4].

Check early enough that you still have room to act. A pre-submission check is only useful if it happens before the deadline, ideally two to three weeks out, so that any genuine overlap can be rewritten and re-verified rather than rushed on the final night [4]. When you re-check, use the same exclusion settings your institution uses, and compare the match list against the previous version rather than comparing the headline percentages alone [4].

Finally, distinguish between what you can fix yourself and what you should raise with your supervisor. Correctly cited quotation overlap in a literature review is usually a conversation about formatting, not misconduct, and supervisors would far rather answer that question early than discover it during a viva [2]. If a large match points to a source you genuinely overlooked, disclosing and correcting it before submission is treated very differently from having it discovered afterwards [4].


Reading your own number is only half the job — the other half is seeing the report exactly as your examiner will, with the match list, the per-source breakdown, and the AI indicator side by side. That is what turnitin0 provides: a pre-submission check that returns the same Turnitin AI and similarity reports your university sees, without adding your thesis to the student paper database.

FAQ

Is 20% plagiarism acceptable for a thesis?
At postgraduate level, 20% is usually above the tolerated band and will typically prompt a written explanation or a review by your department. Whether it is a problem depends entirely on the match list — 20% of correctly cited quotations is a formatting discussion, while 20% concentrated in one uncited source is not [1].

Does the plagiarism percentage include my bibliography and quotations?
It can, depending on the exclusion settings used for that check. Turnitin lets institutions exclude quotes, the bibliography, and small matches, and those settings can move a thesis from the twenties into single digits without any change to the text [3].

Why did my similarity score go up when I only edited a few sentences?
Two common causes are a growing comparison database and self-matching against your own earlier submission. If the previous draft was checked through a repository-based system, it is now in the student paper database and can be matched on a re-check [4].

Is the similarity score the same as the AI writing score?
No. Similarity measures overlap with existing text, while the AI writing indicator estimates machine-generated content and is shown as an asterisk when confidence is below Turnitin's threshold. They are independent metrics reported on the same document [3].

Can I lower my thesis percentage legitimately?
Yes. Shortening verbatim quotations, paraphrasing at the level of argument, adding proper in-text citations, and correcting patchwriting reduce the score without altering your findings. Check with the same exclusion settings your institution applies so the figure you see matches the one your examiner will see [4].

Sources

  1. Interpreting the Similarity Report — https://guides.turnitin.com/hc/en-us/articles/28477544839821-Interpreting-the-Similarity-Report
  2. What Is a Good Similarity Score? — https://www.turnitin.com/blog/what-is-a-good-similarity-score
  3. Viewing the Similarity Report — https://guides.turnitin.com/hc/en-us/articles/22774058814093-Viewing-the-Similarity-Report
  4. How to Reduce Your Similarity Score — https://www.turnitin.com/blog/how-to-reduce-similarity-score

Related articles

Contact us

Email us or reach us on WhatsApp. We typically reply within business hours.