Direct answer
Turnitin plagiarism detection does not decide whether you plagiarised; it compares your submission against three repositories — student papers, internet pages, and publications — and reports the percentage of your text that matches them, with every match traced to a numbered source [1]. The output is called a Similarity Report, and Turnitin states plainly that the percentage is not a plagiarism verdict [2]. Understanding that distinction is the difference between reading your report correctly and panicking over a number that may be entirely legitimate [1].
What does Turnitin plagiarism detection actually check, and what does the Similarity Report show?
Turnitin runs your document through a text-matching algorithm against three distinct collections: current and archived student papers submitted to the repository, billions of indexed web pages, and a periodicals and publications database [2]. It does not query a single "plagiarism database" — it queries all three at once and then separates the hits so you can see where each one came from [2]. That three-part structure is why the same sentence can appear as a web match in one paper and a student-paper match in another.
The result is a Similarity Report in which every matched string is highlighted and numbered, and each number links to a source card showing the originating document, the matched excerpt, and that source's contribution to the total [1]. The headline figure — the Similarity Index — is simply the proportion of your submission's text that matched something in those repositories [1]. It is a measurement of overlap, not a moral judgment, and Turnitin explicitly frames the report as an investigative tool for instructors rather than an automated accusation [2].
Because overlap has many innocent causes, the report also lets instructors exclude quoted material, the bibliography, and matches below a size threshold so the remaining figure reflects substantive unattributed borrowing [1]. Properly cited quotations, standard academic phrasing, and common discipline terminology can all register as matches while remaining completely acceptable [2]. Reading the source list is therefore far more informative than reading the percentage alone [2].
What similarity percentage is considered plagiarism, and why is a high score not automatically misconduct?
There is no universal plagiarism threshold, and no percentage at which Turnitin automatically reports misconduct [3]. Institutions and individual instructors set their own expectations, and many treat a high score as a prompt to open the report and read the sources rather than as a penalty trigger [3]. A submission with a large block of correctly quoted and cited material can legitimately score high, while a paper built from lightly paraphrased unindexed sources can score low [3].
The reverse is equally important: a low similarity score does not prove originality [3]. Because the algorithm only matches against what it has indexed, text paraphrased closely enough to escape exact-string matching, translated material, or content from sources outside the repositories may never appear [3]. That is why experienced markers read the score together with the source list, the citation style, and the assignment brief [3].
The practical rule is to treat the percentage as a navigation aid and the source cards as the evidence [3]. If a flagged passage is a quotation you cited, it is a formatting or citation-density issue, not plagiarism. If a flagged passage is unattributed prose lifted from a source, that is the real problem — and it is fixable before the deadline [3].
Can I check my own draft for Turnitin similarity and AI detection before I submit it?
Sometimes yes, through your institution: many instructors enable a draft or non-repository submission option so students can view a Similarity Report before the final deadline [4]. When that setting is active, the draft is typically checked without being added to the student paper database, so you can revise and resubmit without polluting the repository [4]. Availability depends entirely on the instructor's assignment configuration, so the first step is to check your submission portal rather than assume [4].
Where the draft check is not enabled, students commonly turn to an independent pre-check service to see the same style of report before submitting [4]. Turnitin0 is an independent service and is not affiliated with Turnitin, LLC; it returns a Turnitin AI detection report and a similarity/plagiarism report together, generated without adding your file to Turnitin's student paper database. Documents are English-only, must be over 300 and under 30,000 words, and under 20 MB, with turnaround under 15 minutes in 98% of cases. Pricing is pay-per-use — $3.80 for a single check, or $2.75 per check in a 10-check pack — with no subscription.
The value of a pre-check is diagnostic, not cosmetic [4]. Seeing your own source list lets you add missing citations, convert over-long quotations into paraphrase, and confirm that your bibliography is being excluded as expected [4]. Reviewing the report before the deadline turns a frightening number into a short, concrete revision list [4].
If you would rather walk into submission day knowing exactly what your similarity and AI figures look like, turnitin0 makes that a fifteen-minute step instead of a gamble.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does Turnitin accuse students of plagiarism?
No. Turnitin produces a similarity measurement and a source list; the judgment about whether that overlap constitutes plagiarism is made by a human — your instructor or an academic integrity panel [2]. The report is designed to give them evidence to review, not to issue a verdict [1].
Is a 20% similarity score bad?
Not inherently. Twenty percent could be a properly cited quotation block, a heavily referenced literature review, or genuine unattributed copying — the number alone cannot tell you which [3]. Open the source cards and check whether each match is cited [1].
Why did my similarity score change between two submissions?
The repositories grow continuously as new student papers and web pages are indexed, so the same text can match more sources over time [2]. Instructors can also apply different exclusion settings — quotes, bibliography, small matches — which changes the reported figure [1].
Can I submit my own paper to check it without it counting as my real submission?
Yes, if your instructor enables a draft or non-repository option, the check runs without adding your file to the student paper database [4]. If that option is unavailable, an independent pre-check service can show you an equivalent report before the deadline.
Does Turnitin also detect AI-written text?
Yes — Turnitin's AI writing detection is reported separately from the similarity score, and the two figures measure different things [2]. A paper can have a low similarity score and still carry a high AI-writing indicator, which is why checking both before submission is worthwhile [4].