Direct answer
There is no official acceptable percentage. Turnitin does not set a pass mark, and it does not measure "plagiarism" at all — it measures the percentage of your text that matches other sources, which is why two identical scores can mean completely different things [1]. Most universities that publish a guideline place the working ceiling somewhere between 10% and 25% similarity, but that figure is an institutional convention layered on top of the report, not a rule built into the software [1]. The number that matters far more is which sources are matching and whether the matched text is properly quoted and cited [1].
What similarity percentage does Turnitin consider acceptable?
Turnitin itself publishes no threshold, and its documentation is explicit that the similarity score is not a judgement about plagiarism [2]. The score is simply the proportion of a submission that overlaps with the sources in Turnitin's comparison databases, after whatever exclusions the instructor has switched on [2]. Because those settings are configurable, the same essay can return different scores in two different courses — one marker may exclude quotations and the bibliography while another includes them [2]. That is why Turnitin's own guidance tells instructors to read the score alongside the match overview rather than in isolation [2].
Where a numeric ceiling does exist, it comes from the institution. Departments commonly describe bands informally — under roughly 10% is usually treated as unremarkable, 10–25% invites a closer look at the sources, and anything above about 25% is likely to trigger a conversation — but these are local conventions, not Turnitin policy [2]. Turnitin's help documentation also notes that small matches are excluded by default and that the score depends on which databases were searched at the time of the check, so a score is a snapshot of one configuration rather than an absolute property of your writing [2]. If your handbook states a figure, that figure governs your submission; if it states none, the source list governs.
The practical consequence is that chasing a target number is the wrong optimisation. A 9% score built entirely from unattributed paragraphs lifted from one paper is a serious integrity problem, while a 22% score made up of correctly cited quotations, a reference list and a handful of generic academic phrases is normally unremarkable [2]. Read the percentage as a triage signal that tells you where to look, and read the sources as the evidence [2].
Why can a high similarity score still be innocent, and what does Turnitin's report actually measure?
A high score is frequently innocent because the report counts matches, not misconduct. Titles, module names, standard definitions, quoted legislation, interview transcripts and reference lists all generate matches against sources that were legitimately used [3]. Turnitin's own commentary on report interpretation stresses that the match overview should be reviewed match by match, since a single long match to one source is a very different signal from a scattering of short matches across dozens of sources [3]. A score of 30% concentrated in a bibliography that was not excluded looks alarming and means nothing.
Repeated submission inflates scores in a way students rarely anticipate. If you have already run your draft through a checker that stores papers in a student repository, your own earlier version can become a match source on the next run, and so can a group member's near-identical draft [3]. Turnitin's documentation also warns that matches to other students' papers are among the most common causes of unexpectedly high percentages, and that these are visible in the source list as named student papers rather than publications [3]. That is one reason a non-repository check is the more useful diagnostic before submission.
The report also measures similarity of wording, not originality of thought. Paraphrase that keeps the original sentence architecture can still match, and a well-argued original paragraph that happens to reuse a common framing can match too [3]. Turnitin's guidance is to treat the report as a starting point for reviewing your own citations, not as a verdict, and to check whether each highlighted passage is quoted, cited and genuinely integrated into your argument [3]. If every match is accounted for, the percentage is a description of your source base rather than a problem.
How can I check my own similarity and AI scores before I submit?
The reliable way to know what your marker will see is to run the same kind of check they will run, on the same file, before the deadline. Turnitin0 provides a non-repository pre-submission check that returns two PDFs in one order — a Turnitin AI detection report and a similarity report — formatted as the reports appear in an institutional LMS [4]. Because the file is not added to Turnitin's student paper database, your own draft cannot come back as a match against itself on a later run, which removes one of the most common causes of a misleadingly high score [4]. English documents between 300 and 30,000 words and under 20 MB are supported, and most orders complete in 5–15 minutes [4].
Reading the output correctly matters as much as obtaining it. On the AI side, Turnitin displays an asterisk (*%) instead of an exact figure when its detection confidence falls below the 20% threshold, so a starred result is a low-confidence signal rather than a hidden high score [4]. On the similarity side, open the match overview and check each source individually: quoted and cited material, references and common phrases should be discounted, and what remains is the portion you genuinely need to rewrite or attribute [4]. Turnitin0 has delivered over 100,000 reports to more than 20,000 students, and the service is independent of and not affiliated with Turnitin, LLC [4].
If the remaining unattributed portion is large, the fix is editorial rather than cosmetic: rewrite the flagged passages in your own structure, add the missing citations, and re-check the revised file rather than guessing at the effect [4]. Students who do this routinely find that the score they were worried about was almost entirely references and quotations, and the small residue is a short, specific list of paragraphs to rework [4].
Guessing what your marker's report will look like is the part you can eliminate entirely — and turnitin0 exists precisely so you can see the real AI and similarity reports, with the score, the flags and the matched sources laid out exactly as your institution will read them, before anything is submitted.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is 20% plagiarism acceptable in Turnitin?
Turnitin sets no threshold, so "acceptable" is defined by your institution, not the software [1]. A 20% similarity score is common and often harmless when the matches are quotations, references and standard academic phrasing, but it needs a match-by-match review before you decide [3].
Is a 0% similarity score good?
Not necessarily. A 0% score can mean excellent original writing, but it can also mean the check ran against a limited database or that your citations were excluded from the count [2]. Turnitin advises reading the source list rather than treating a low number as proof of quality [2].
Does Turnitin detect plagiarism automatically?
No. Turnitin produces a similarity report identifying matching text; the determination of plagiarism is made by a human who judges whether the matching text was properly attributed [1]. That is why the same score can lead to different outcomes in different departments [2].
Why is my score higher than my friend's when we cited the same sources?
Exclusion settings, database coverage and submission history differ between checks, and a stored earlier draft can itself become a match source [3]. Running a non-repository check on your final file gives you a score that reflects your text rather than your submission history [4].
What should I do if my similarity score is above my department's limit?
Open the match overview, discount properly cited and quoted material, and rewrite or attribute the remainder in your own words [3]. Then re-check the revised file so you know the new number before the deadline rather than after it [4].