Direct answer
To check content for plagiarism in Turnitin, you submit a file (usually .docx, .pdf, or .txt) and Turnitin compares its text against its student paper repository, the live and archived internet, and periodical and publication databases, then returns a Similarity Report with every matched passage highlighted and linked to its source [1]. The percentage you see is the share of text that matched something else — it is not a plagiarism verdict, and Turnitin itself does not decide whether you plagiarised [1][3]. Students on institutional accounts can only run that check when their instructor has enabled submissions and report access, which is why many writers preview the report on a draft before the real deadline [2][4].
How does Turnitin check content for plagiarism, and what does the similarity report actually show?
Turnitin works by text matching rather than by judgement. When a file is processed, the system breaks the submission into segments and compares them against three source pools: the student paper repository (work previously submitted to institutions using Turnitin), the internet including archived pages, and a licensed collection of journals, periodicals, and publications [1]. Anything that overlaps is marked, and the report links that passage back to the specific source it matched, so you can see exactly which sentence came from where rather than staring at a single number [1].
The Similarity Report is organised into layers. The Match Overview gives you the highest-matching sources in descending order, and the All Sources view lists every match found, including overlaps that were filtered out of the headline percentage [2]. Each match is colour-coded and, when you open it, shows the source alongside your own text so you can judge whether the overlap is a quotation, a paraphrase, a common phrase, or genuine copying [1][2].
Three settings change what the number means, and they are the most common source of confusion. Quotation marks and bibliography entries can be excluded from the index, small matches below a word-count threshold can be filtered out, and overlapping matches from the same student paper can be collapsed into one [2]. Two students can therefore submit identical files and see different percentages purely because their instructors configured the report differently — the underlying matches are the same, the arithmetic is not.
What the report does not show is intent. Turnitin's own documentation is explicit that the similarity index measures matched text, not academic misconduct, and that the same percentage can mean very different things depending on which sources matched [1][3]. A report is evidence for a human to read, not a ruling, so the useful question is never "what is my score" but "which passages matched, and can I defend each one" [1].
What similarity percentage is considered acceptable, and what do the matched sources mean?
There is no universal pass mark. Turnitin does not publish a threshold at which a submission becomes plagiarism, and institutions set their own expectations in their academic integrity policies [3]. A figure that raises no questions in a lab report with standard method wording can be alarming in an essay that quotes heavily, and the reverse is also true — a low percentage can still hide one copied paragraph if the rest of the document is original [3].
What matters far more than the headline number is the composition of the matches. Matches to journals and publications, especially long unbroken runs of highlighted text, usually indicate material that needed a citation or a rewrite [1][3]. Matches to a single student paper are worth investigating carefully, because that pattern is what a shared-notes or contract-cheating case looks like. Matches scattered across many small sources and common academic phrasing ("the results suggest that", "in recent years") are typically noise, and instructors read them that way [3].
A high percentage driven by properly cited quotations is not misconduct, but it is still a writing problem worth fixing. Examiners read a report alongside the text, so an essay that is 40% quoted may be fully referenced and still fail on originality of argument [3]. The practical move is to reduce reliance on long quotations, paraphrase where the idea matters more than the wording, and make sure every retained quotation carries a citation that the report can see [1][2].
Finally, treat the percentage as a diagnostic, not a target. Chasing a specific number by paraphrasing mechanically tends to produce text that reads worse and still matches, because synonym-swapping leaves the sentence structure intact [3]. Read the matched sources one by one, decide for each whether it is correctly attributed, and fix only the ones that are not — that is the workflow the report was designed to support [1][3].
Can I check my own draft for plagiarism before submitting it and get the same report my instructor sees?
Usually not through your institution. On a standard university account, whether you can submit at all, whether you can see the report, and whether you may resubmit are all controlled by the instructor's assignment settings [2][4]. Many courses leave student report access switched off until after the due date, and where self-submission is allowed, the draft may be added to the institutional paper repository — which can create a match against your own later submission [4]. That is the practical reason students look for an independent route to preview the report.
An independent check is only useful if it reproduces the same report format your instructor will read: the same match highlighting, the same source links, the same similarity index, plus the AI writing indicator that increasingly sits alongside it. A checker that returns its own proprietary percentage tells you little about what your marker will see. What you want is the actual report output — cover page, similarity summary, flagged passages with sources — so you can fix the specific passages before the real submission [1][2].
When you do preview, work from the sources rather than the score. Open each match, confirm the citation is present and correctly formatted, and rewrite any passage where the wording tracks the source too closely even if the idea is attributed [1]. If your draft was drafted or polished with an AI assistant, run the AI writing check in the same pass, because a clean similarity report does not mean a clean AI report — the two indicators measure different things and instructors now look at both [4].
One practical caution: never submit your final version to a repository-based checker you do not control. If the draft enters a student paper database, a later legitimate submission of the same text can match itself. Use a non-repository check for drafts, keep your own copies, and treat any preview as a rehearsal for the report you will actually receive [4].
If you would rather see the real thing than guess, turnitin0 lets you upload your draft and get the same Turnitin AI and similarity reports your instructor sees — matched sources, highlighted passages, and the AI indicator in one delivery, checked without adding your file to the student paper database.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does Turnitin check my content against the whole internet?
Yes — the comparison pool includes current and archived internet pages, periodicals and publications, and the student paper repository, and each match is linked to its specific source in the report [1].
Is a 20% similarity score bad?
Not inherently. Twenty percent can be entirely cited quotations or standard method wording, or it can contain one copied paragraph; the score only becomes meaningful when you read which sources matched and how long the matched runs are [3].
Can I see my similarity report before the deadline?
Only if your instructor has enabled student report access for that assignment, since visibility and resubmission rights are set per assignment rather than by the student [2][4].
Does a clean similarity report mean my AI score is fine too?
No. Similarity measures text overlap with existing sources, while the AI writing indicator estimates machine-generated phrasing — they are separate signals and a draft can pass one and fail the other [4].
Will checking my draft get it added to Turnitin's student paper database?
It depends entirely on the service. Institutional self-submissions often are added to the repository, so for drafts use a check that explicitly excludes repository storage [4].