Direct answer
There is no single universal percentage that is "acceptable" for plagiarism. Turnitin's similarity score is a measure of matched text, not a plagiarism verdict, and each university sets its own threshold [1]. In practice, many institutions treat roughly 10–24% as unremarkable for coursework that includes quotations and a reference list, while scores above about 25% usually trigger a closer look rather than an automatic penalty [2]. The number only becomes meaningful when you know what matched, why it matched, and what your own department's policy says.
What Similarity Percentage Does Turnitin Actually Flag, and What Do Universities Treat as Acceptable?
Turnitin does not "flag" a paper at a fixed number. The similarity score is simply the proportion of a submission that matches text in its comparison database, and the same percentage can mean very different things in a literature review, a lab report, or a reflective essay [2]. Because the score is descriptive rather than judgmental, a high number is best read as an instruction to look at the match breakdown, not as a verdict on academic integrity [1].
Where a line does exist, it is drawn by your institution, not by the software. Many universities publish banded guidance rather than a hard cut-off: scores in the low teens are often considered normal for referenced work, mid-range scores prompt a marker to inspect the highlighted passages, and only sustained unattributed copying leads to misconduct procedures [2]. A department may also apply stricter limits to a short reflective piece than to a 5,000-word dissertation, so the same 18% can be routine in one context and concerning in another.
The practical takeaway is to treat the percentage as the start of a conversation with your own marking criteria. Check your module handbook or academic integrity page for a stated threshold, and if none is published, assume the expectation is that all borrowed material is properly quoted, cited, and paraphrased — not that a particular number is safe [1]. Students who understand the score this way avoid both false panic and false confidence.
Does the Type of Matched Source — Quotes, References, Common Phrases — Change Whether a Percentage Is a Problem?
Yes, and this is the single biggest reason a "high" score is often harmless. Turnitin includes quoted text and bibliographic entries in the similarity score unless the instructor has switched on the exclude-quotes and exclude-bibliography options, so a correctly referenced essay can legitimately display a double-digit figure [3]. A block quotation of 200 words plus a reference list of 30 entries can push a score up without a single integrity issue.
The match overview is where you separate real problems from noise. Each highlighted passage is linked to its source, so you can see instantly whether the overlap comes from a properly attributed quotation, a common phrase such as "in recent years," a template phrase from the assignment brief, or an unattributed paragraph lifted from a journal article [3]. Only the last category is a genuine concern; the first three are normal academic writing.
It also helps to know that the match percentage is not additive in the way students assume. Multiple short matches to the same source are grouped, and small fragments below the reporting threshold are ignored, which is why two drafts with similar content can show noticeably different scores [3]. If your score looks inflated, review the highlighted sources first and consider whether your quotations are excessive — trimming long quotes and paraphrasing more tightly often lowers the number honestly.
How Can a Student See Their Own Turnitin Similarity and AI Score Before the Real Submission?
The safest way to answer "is my percentage acceptable?" is to look at the actual report before the graded submission. Many institutions enable a pre-submission or draft-check route, and independent checking services return the same style of similarity report and AI writing report that a marker sees [4]. Seeing the highlighted sources yourself removes the guesswork about which passages are driving the number.
A pre-submission check is diagnostic, not a loophole. You use it to find unattributed overlap, over-long quotations, and accidental copy-paste from your own earlier drafts, then fix them properly with quotation, citation, or rewriting [4]. Because the report separates similarity from AI writing detection, you also learn whether any section is being read as machine-generated — useful if you drafted with ChatGPT, Claude, or Gemini and then edited by hand.
Timing matters as much as the check itself. Reviewing a draft a day or two before the deadline leaves room to paraphrase, add citations, and re-check, whereas a last-minute check only tells you there is a problem you can no longer fix [4]. Treat the percentage as a revision signal: if the flagged passages are all correctly cited, you are fine; if they are not, you still have time to correct them.
If you would rather see the real report on your own draft instead of guessing what your marker will see, turnitin0 gives you the same Turnitin similarity and AI writing reports — with the highlighted sources, score, and flags — before you submit, so you can fix genuine overlap and leave the properly cited material alone.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is 20% similarity bad?
Not automatically. Around 20% is common for coursework with quotations and a reference list, and it only becomes a problem if the highlighted text is unattributed [2]. Open the match overview and check what produced the score before worrying.
Does Turnitin count my reference list in the percentage?
It can. Quoted and bibliographic text is included unless the instructor has enabled the exclude-quotes and exclude-bibliography settings [3]. That is one reason a fully referenced essay can show a higher number than expected.
Can I get in trouble for a high similarity score even if I cited everything?
A properly cited paper with a high score is usually fine, because the report distinguishes matched text from misconduct [1]. Your institution's policy and the marker's reading of the highlighted passages decide the outcome, not the raw percentage.
What is the safest similarity percentage to aim for?
Aim for a draft where every match is either a correctly quoted and cited passage or a common phrase, rather than chasing a specific number [2]. Reviewing your own report before submission is the most reliable way to confirm that.
How do I check my own percentage before submitting?
Use a pre-submission check that returns the same similarity and AI writing report format your marker sees, then revise any unattributed overlap before the deadline [4]. This gives you time to paraphrase, add citations, and re-check.