Direct answer
There is no single "good" plagiarism score, but most universities treat a Turnitin similarity score under roughly 10–15% as unproblematic for a standard essay, with 15–25% needing a closer look and anything above 25% usually requiring explanation or revision. Turnitin itself publishes no universal cut-off, because the similarity score measures matched text — not plagiarism — so the number only becomes meaningful once an instructor reads the underlying matches in context [1]. A "good" score is therefore one where every highlighted match is properly quoted, cited, or a harmless stock phrase.
What Similarity Percentage Does Turnitin Consider Acceptable, and How Do Universities Actually Read It?
Turnitin does not define an acceptable threshold, and it says so plainly: the similarity score is the percentage of a submission's text that matches sources in its database, and a match is not evidence of misconduct [1]. Two papers with the same percentage can carry completely different meaning — one may be quoting a statute at length with correct attribution, the other may have copied a paragraph without a citation. That is why the score is designed as a review prompt, not a verdict, and why the number alone can never tell you whether your work is acceptable [2].
In practice, institutions fill the gap by setting their own expectations, often in academic integrity policy, module handbooks, or marking rubrics [2]. Turnitin's guidance for interpreting the report stresses that the percentage is a starting point for review, and that the source list, the matched passages, and the assignment type all matter more than the headline figure [2]. A literature review built from secondary sources will naturally score higher than a reflective portfolio, and a lab report's methods section may match a template used by every student on the course. Judging a score without knowing the genre is a common student error.
A workable reading of the bands, used widely in UK, US, Canadian, Australian, New Zealand, and Irish universities:
| Similarity band | Typical institutional reading | What to do |
|---|---|---|
| 0–10% | Normal for original writing; matches usually references or stock phrasing | Verify the highlighted strings are cited |
| 10–15% | Usually fine for essays with heavy referencing | Check that no single source dominates |
| 15–25% | Needs review; may be acceptable for source-heavy work | Inspect each match; paraphrase or add citation |
| 25%+ | Usually flagged for discussion with the marker | Revise before submitting, or explain the overlap |
These bands are a practical heuristic, not a rule book — your own programme's policy always overrides them [2]. The safest habit is to look at which sources are matched and how long each matched string is. A 30% score made of twenty short citations from twenty different papers is a very different document from a 30% score where one journal article accounts for 28 points of it.
Why Can a High Similarity Score Appear in Work You Wrote Yourself?
Original writing still produces matches, because Turnitin compares strings of words rather than ideas. Standard academic phrasing — "the results of this study suggest", definitions of established terms, method descriptions, institutional boilerplate — recurs across thousands of papers and gets picked up as a match even when you composed the sentence independently [3]. Long quotations, block quotes, and paraphrases that stay close to the original wording also register, and so do reference lists, table captions, and appendices unless those sections are excluded.
The filters matter enormously here. When quotes, bibliography, and small matches are excluded, a reported percentage can drop by many points without a single word of the body changing [3]. If your score looks alarming, first check whether the report was generated with the filters your department uses — comparing a filtered preview against an unfiltered one is not comparing like with like. Turnitin's own reporting documentation notes that the similarity score and the AI writing indicator are separate signals computed from different things, so a clean similarity score tells you nothing about whether the AI indicator will fire [3].
There is also a mundane cause: submitting the same essay to a previous module, or reusing a passage from your own earlier assignment, produces a match against your own prior submission if it entered the student paper repository. Self-matching is still a match, and many institutions treat unacknowledged reuse of your own work as a separate integrity issue. If a single source accounts for most of your percentage, read the matched text before assuming the worst — the fix is usually citation, paraphrase, or a rewrite of one paragraph, not a rewrite of the paper.
How Can You Check Your Own Similarity and AI Scores Before You Submit?
Most students cannot see a Turnitin report before the deadline, because the preview setting is controlled by the instructor and is frequently switched off [4]. Where it is enabled, you get one draft-level view and a limited number of resubmissions; where it is not, you submit blind. That asymmetry is the reason pre-submission checking exists as a category: the useful moment to discover a 34% similarity score is three days before the deadline, not three days after it [4].
A pre-submission check should reproduce what your marker sees rather than approximate it. Turnitin0 produces the same two documents an instructor receives — a similarity report with the matched sources and highlighted passages, plus an AI writing report — from a .docx, .pdf, or .txt file between 300 and 30,000 words and under 20 MB. The check is non-repository, so your draft is not added to the student paper database and will not match against itself later, and most reports return in under 15 minutes. You get both PDFs in one checkout, which matters because a similarity score you are happy with can sit alongside an AI indicator you are not.
Read the report the way a marker would. Open the source list, sort by percentage of matched text, and inspect the largest contributors first — those are the ones that decide the conversation. Then check the AI report separately, since a passage you paraphrased by hand can still read as machine-generated to a detector. Keep the PDFs; if a score is queried later, a dated pre-submission report showing what you changed is far stronger than a verbal explanation.
Knowing the bands is only half the job — the other half is seeing your own numbers, in the same format your marker will open, while there is still time to act on them. That is exactly what turnitin0.com is built for: a real Turnitin similarity report and AI writing report on your draft, delivered in minutes, with no subscription and nothing added to the student paper database.
※ Turnitin0.com - Actual [Turnitin AI](https://www.turnitin0.com/) Report Cover, Score, Flag And Similarity Summary
FAQ
Is 20% a bad Turnitin score?
Not automatically. Turnitin publishes no cut-off, and 20% is often made up of reference lists, quotations, and standard academic phrasing [1]. It becomes a problem only if a single source dominates the matches or a copied passage lacks citation [2].
Does a 0% similarity score mean my work is original?
No. A low score can mean the sources you used are not in Turnitin's database, or that your paraphrasing drifted far from the original wording [1]. Originality is judged on attribution and argument, not on the absence of matches.
Why did my score rise after I added references?
Reference lists and bibliographies are matched text unless the report excludes them [3]. If your department's view filters bibliography out, your score will read lower than an unfiltered preview of the same file.
Can I see my similarity score before submitting?
Only if your instructor enables the preview setting, which many do not [4]. A pre-submission check on turnitin0.com gives you the same similarity and AI reports without adding your draft to the student paper database.
Does a good similarity score mean the AI detector will pass me?
They are separate signals. The similarity score compares text to sources, while the AI writing indicator estimates how much qualifying text looks machine-generated [3]. A draft can score 8% similarity and still carry a high AI percentage, so check both reports.