Direct answer
A similarity check is an automated comparison of your document against a large body of existing text — web pages, journal articles, and previously submitted student papers — that produces a percentage showing how much of your writing matches other sources. It is a text-matching measurement, not a plagiarism verdict: the percentage tells you how much overlap exists, while the report tells you where that overlap sits and whether it is properly attributed [1]. Most universities run this check through Turnitin, and the resulting Similarity Report is what an instructor reads alongside your submission [1].
What Does a Similarity Check Actually Compare, and What Does the Percentage Mean?
A similarity check works by breaking your document into text segments and searching those segments against three broad pools of material: the open web, subscription academic databases and journals, and a repository of papers previously submitted by students [2]. Every segment that matches something in those pools is flagged, colour-coded, and listed as a source in the report. The headline number is simply the proportion of your document's text that was matched — nothing more, and nothing less [2].
That percentage is a sum, and sums can mislead. When two or more sources overlap with the same passage, Turnitin counts the passage more than once in the total, so a paper can show a higher score than the amount of genuinely duplicated text would suggest [2]. Direct quotations, reference lists, and common phrases all match text by design, which is why the report lets you exclude quotes and bibliography and see a recalculated figure [2]. Reading the score without the underlying source list is the single most common way students misjudge their own work [1].
There is also no universal "acceptable" threshold. Turnitin publishes score bands as a rough guide rather than a rule, because a literature review that quotes heavily and cites correctly can legitimately score far higher than a short original essay [2]. What matters is the pattern: one large block matching a single source is a very different signal from scattered 1% matches across a dozen references [1].
Is a High Similarity Score the Same Thing as Plagiarism?
No. Plagiarism is the use of someone else's words, ideas, or data without proper attribution, and it is defined by institutional policy — not by a percentage [3]. A correctly cited quotation will always match text in a similarity check, because the words genuinely belong to another author; that is attribution working as intended, not misconduct [3]. Conversely, a paper can score 4% and still be plagiarised, for example when a passage is paraphrased closely without a citation or when work was produced by someone else entirely [3].
This is why the report and the score must be read together. The report shows you the matched source, the matched passage, and the surrounding context, which is exactly the information an academic integrity officer would look at [1]. If a flagged passage is quoted and referenced, the match is expected; if it is unquoted and uncited, the same match is a problem [3].
A further complication is that a similarity check and an AI writing check are two separate measurements inside the same report. A similarity score says nothing about whether text was generated by a language model, and an AI score says nothing about whether sources were attributed [2]. Students who assume one number covers both often misread their own results.
How Can I Run a Similarity Check on My Own Draft Before I Submit It?
In most institutions, you cannot simply log into Turnitin and check your own draft. Student accounts only see submissions made through an assignment the instructor has created, and whether you can view the resulting report at all depends on settings the instructor controls [4]. Some departments deliberately enable draft or non-repository submissions so students can test work early, but this is a course-by-course decision rather than a standard feature [4].
Where that option is not available, students commonly use an independent pre-submission check to preview what a similarity report will look like on their own file. This is a preview, not a substitute for your institution's submission: it lets you see which passages match, which sources are involved, and how much of the percentage comes from quotations and references you have already cited [1]. Seeing the source list before the deadline is what turns an abstract percentage into an actionable edit list.
Two practical habits make that preview useful. First, check the version of the file you actually intend to submit, including headings, tables, and the reference list, because formatting changes alter how text is segmented and matched [2]. Second, compare the score against the source list rather than against a target number, and fix any unquoted, uncited match you find — that is the only part of the report that carries real risk [3].
Reading a similarity report is far easier once you have seen a real one attached to your own draft, with the score, the flagged passages, and the source list all in front of you. turnitin0 produces exactly that preview: upload your.docx,.pdf, or.txt file and you receive a Turnitin AI detection report and a similarity report as downloadable PDFs, formatted the way your instructor's system presents them. Most orders complete in under 15 minutes, files are checked without being added to the student paper database, and 100,000+ reports have been delivered to 20,000+ students so far.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is a similarity check the same as a plagiarism check?
No. A similarity check measures text overlap; plagiarism is an attribution and policy judgement made by your institution [3]. The report is the evidence an instructor uses, not the verdict itself [1].
What similarity percentage is safe to submit?
There is no universal safe number, and Turnitin explicitly treats its score bands as guidance rather than a rule [2]. Read the source list instead: cited quotations and references are expected matches, while unquoted, uncited passages need rewriting [1].
Why is my similarity score higher than the amount I actually copied?
Overlapping sources can be counted more than once in the total, and quotations plus reference lists match text by design [2]. Excluding quotes and bibliography in the report shows a recalculated figure that is usually closer to the real overlap [2].
Can I check my own draft in Turnitin before submitting?
Usually not directly — student accounts only see submissions made through an instructor-created assignment, and report visibility is controlled by the instructor [4]. Some courses enable draft or non-repository submissions, but it varies by module [4].
Does a similarity check also detect AI-generated writing?
No. Similarity and AI writing detection are separate measurements, though modern reports present both [2]. A low similarity score does not mean the text will pass an AI writing check.