Direct answer
There is no official similarity percentage that turns into plagiarism. Turnitin does not publish a cutoff, and universities set their own limits, so a score of 18% is not automatically plagiarism and a score of 4% is not automatically safe [1]. The similarity score measures how much of your text matches sources in a database, not whether you plagiarised [2]. What decides the outcome is the match breakdown: whether the flagged text is quoted, cited, and correctly attributed, or copied and passed off as your own.
Is There a Specific Similarity Percentage That Counts as Plagiarism?
No. Turnitin is explicit that the similarity score is not a plagiarism score, and it publishes no threshold at which a submission becomes plagiarism [2]. The percentage simply reports how much of a document matched text already stored in the comparison database, which includes journal articles, web pages, and previously submitted student papers. Because the number describes overlap rather than intent, the same 22% can be entirely innocent in one essay and serious misconduct in another [2].
Institutions, not Turnitin, decide what happens after that number appears. Many departments publish informal guidance in the region of 10% to 25% excluding references and quotations, but those figures are local policy conventions rather than a product rule, and they vary by discipline, assessment type, and level of study [2]. A law school's tolerance for quoted statute and a computer science lab report's tolerance for shared code are not comparable, so importing a number from a forum thread or a different university is unreliable.
The practical consequence is that you cannot answer the question with a single figure, and neither can your instructor without opening the report. Turnitin's own documentation directs markers to the match overview, where every source is listed with its own contribution, precisely because the headline percentage hides the detail that matters [2]. A submission whose 30% comes from one correctly cited block quotation is read very differently from one whose 30% is scattered across eight unattributed paragraphs.
Why Can a High Score Be Fine and a Low Score Still Be Plagiarism?
High scores are frequently legitimate. Reference lists, in-text citations, standard methodology wording, subject-specific terminology, and common academic phrases all match existing text by design, and Turnitin counts them [3]. Students who reuse material from their own earlier submission can also see a large match, because their previous paper sits in the repository [3]. None of that is misconduct, and much of it is exactly what a well-referenced essay should look like.
Low scores prove nothing in the other direction. A single copied paragraph in a 4,000-word essay may register only two or three percent, yet it can be the most serious problem in the document [3]. Turnitin's guidance on interpreting results stresses that the percentage must be read together with what matched and how it was used, which is why markers look at the highlighted passages and the source list rather than the number alone [3]. A low score is a weak signal, not a clearance.
The same logic applies to paraphrasing. Close paraphrasing that keeps the original sentence structure while swapping synonyms may match only partially, or not at all, and still breach attribution rules. Conversely, a correctly quoted and cited sentence will be flagged as similarity and is not plagiarism [3]. The question to ask about each highlight is not "how big is this percentage" but "would a reader know this idea came from someone else."
How Can a Student See Their Real Similarity Score and Match Breakdown Before Submitting?
The only reliable way to know what your score means is to read the report the way your marker will. A similarity report shows the overall percentage at the top and, below it, a match overview listing every source with its individual contribution, plus the highlighted text in context [4]. That breakdown is what converts a meaningless number into a judgement: you can see whether a match sits inside quotation marks with a citation attached, or whether it is unattributed prose that needs rewriting.
Access is the usual obstacle. Some institutions enable draft or self-check submissions that let students view a report before the graded deadline, but many do not, and where they do not, the report only becomes visible after the real submission has been made [4]. Students in that position are effectively asked to guess how their draft will be read, which is why so many search for a percentage threshold in the first place.
If you do have access, work through the matches in descending order of contribution rather than reacting to the total. For each one, check whether the source is cited, whether quoted wording is inside quotation marks, and whether the surrounding sentences genuinely restate the idea in your own words [4]. Fix the matches that fail those tests, and the total will usually fall on its own as a side effect rather than as the goal.
Reading your own draft through a marker's eyes is far easier when you can see the actual report before the deadline does the deciding for you. That is exactly what turnitin0 gives you: a real Turnitin similarity report plus the AI writing report, delivered together so you can inspect every match, every source, and every flag while there is still time to revise.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Does Turnitin flag plagiarism at a certain percentage?
No. Turnitin reports similarity and does not classify a submission as plagiarism at any percentage; its documentation states there is no magic number, and interpretation is left to the institution [2].
Is 20% similarity bad?
Not necessarily. Twenty percent made up of references, quoted material, and standard terminology is normal, while twenty percent of unattributed copied prose is a problem — the match overview shows which you have [1].
What similarity score do universities usually accept?
Many departments work informally between roughly 10% and 25% excluding quotations and references, but these are local conventions rather than a universal rule and vary widely by discipline [2].
Can I be accused of plagiarism with a low similarity score?
Yes. A single uncited paragraph in a long essay may add only a few percent, yet it can still be the most serious issue in the document [3].
How do I check my similarity before submitting?
Some universities enable draft self-checks, but where they do not, the report only appears after submission [4]. Students can instead review a real similarity report on a draft through an independent pre-submission check and fix matches while there is still time.