Direct answer
Turnitin does not set any percentage that is considered plagiarism. There is no 10%, 15%, 20%, or 30% cut-off built into the Similarity Report, and Turnitin states that the similarity score is an indicator rather than a judgment of misconduct [1]. Whether a score is treated as plagiarism is decided by your instructor or institution, using their own academic-integrity policy and the context of the matched text [1]. A high similarity score can be entirely legitimate, and a low score does not prove originality.
Does Turnitin Set a Fixed Similarity Percentage That Counts as Plagiarism?
No. Turnitin is explicit that it does not define plagiarism by a number, and the company positions the Similarity Report as a tool for interpretation rather than a verdict [2]. The percentage you see is simply the proportion of your document that matched text in the Turnitin repositories it was compared against — nothing more and nothing less [2]. Turnitin does not label a submission "plagiarised" at any threshold, and it does not attach a pass/fail band to the score [2].
This is the single most misunderstood point about the tool. Students often assume there is an official Turnitin rule — something like "under 20% is fine" — but no such rule exists in the product itself [1]. The number is descriptive, not evaluative. It tells you how much text matched; it says nothing about why it matched, whether it was cited, or whether your instructor will care [1].
The other half of the misconception is that a low score equals safety. A paper assembled from paraphrased sources with no citations can score surprisingly low while still breaching academic-integrity rules, and a heavily quoted literature review can score high while being perfectly compliant [1]. Turnitin's own guidance warns against reading the score as a proxy for integrity in either direction [2].
How Do Universities Actually Decide Whether a Similarity Score Is Too High?
Institutions assess the substance of the matches, not the headline number. Academic-integrity policies typically direct staff to consider whether the matched material was properly attributed, how extensive and continuous the copied passages are, whether the duplication sits in the analysis or only in quotations and references, and whether there is evidence of intent [3]. Two submissions with an identical 24% score can produce completely different outcomes under that test [3].
Most institutions also configure the report before interpreting it, which is why your score may differ from a friend's on a similar draft. Reviewers commonly exclude quotations, the bibliography, and small matches below a set word count, and some exclude all sources under a percentage floor [2]. Those settings can move a score by many points without a single word of the essay changing [2]. This is also why comparing your percentage to a classmate's is close to meaningless unless you know both reports used the same filters.
Where a case does proceed, it usually rests on the match breakdown rather than the total. An examiner looking at a 31% report will open the match overview and see whether the overlap is one long unattributed block from a single source, or forty small fragments of technical terminology scattered across the document [3]. The first pattern is a problem; the second frequently is not. Departments that publish guidance tend to phrase thresholds as triggers for review rather than definitions of offence — a score above a certain level may prompt a conversation, but the conversation, not the number, determines the outcome [3].
How Can You See Your Own Similarity Percentage and Match Breakdown Before Submitting?
The only reliable way to know how your draft will be read is to look at the same report your instructor will look at — the score plus the underlying matches. Inside Turnitin, the Similarity Report opens with a match overview that lists every source matched, the percentage each one contributed, and the exact passages highlighted in your text [4]. That per-source breakdown is what turns a single alarming number into an actionable list, because it shows you which matches are quotations, which are references, and which are genuinely unattributed prose [4].
When you review your own report, work through it in a fixed order. First, check the filters applied to the report, since an unfiltered score is almost always higher than the one your department uses [4]. Second, open the largest single match and read the highlighted passage in context — if it is a direct quotation with a citation, it is doing its job; if it is a paraphrase that drifted too close to the original, it needs rewriting [4]. Third, look at the reference list and bibliography matches, which inflate totals without any integrity issue at all [2].
The gap in that workflow is access. Turnitin's Similarity Report is generated inside an institutional course, so most students cannot run their own draft through it before the deadline — they either submit blind or rely on free checkers that use different databases and produce numbers that do not correspond to the real report. That is the specific problem a pre-submission check solves: you get the actual report format, with the score and the match breakdown, while there is still time to revise the passages that need it.
Reading a similarity percentage only becomes useful once you can see the report behind it — the score, the source list, and the highlighted matches side by side. That is exactly what turnitin0 provides: a pre-submission check that returns a Turnitin AI detection report and a similarity report together, in the same format your professor sees, so you can judge your own matches before anyone else does.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is 20% similarity bad on Turnitin?
Not inherently. Twenty percent is neither a Turnitin threshold nor a universal warning sign; Turnitin attaches no meaning to the figure itself [1]. Whether 20% matters depends on what the matched 20% consists of — quotations and references are usually excluded by the reviewer's filters, while unattributed prose is a problem at any percentage [2].
Is 10% plagiarism safe?
No score is "safe" by definition, because the assessment looks at attribution rather than volume. A 10% report containing one uncited paragraph copied verbatim can be treated as plagiarism, while a 40% report of properly cited quotations often is not [3]. Read the match breakdown, not just the total.
What similarity score do most universities treat as too high?
There is no shared figure. Institutions differ, and many deliberately avoid publishing a number, instead treating a high score as a prompt for review rather than an automatic finding [3]. If your department publishes guidance, that document — not a general benchmark — is the one that applies to you.
Why is my similarity score higher than my friend's for similar work?
Report filters are the usual reason. Exclusions for quotations, bibliography, and small matches can shift a score substantially, and different modules or institutions configure them differently [4]. Two reports are only comparable when the same exclusions are applied [2].
Does Turnitin detect AI writing with the same percentage?
No. The AI writing indicator is a separate metric from the similarity score and is reported independently, with Turnitin displaying an asterisk rather than an exact figure when its confidence is below its threshold [1]. A clean similarity score tells you nothing about the AI indicator.