A good Turnitin similarity score is generally considered to be under about 15-20%, provided that every matched passage is properly quoted, paraphrased and cited — because the figure measures matching text against Turnitin's databases, not the “amount you plagiarised.” There is no single universal pass mark: what counts as an acceptable plagiarism percentage depends on your university, your assignment type and, above all, whether your matches are legitimately referenced.
This guide explains what a good similarity score actually is, what the percentage really means, how the colour bands are read, what inflates a score harmlessly, and how to legitimately lower it before you submit — with a quick-reference table and a worked example you can copy.
Is there a universally “good” Turnitin similarity score?
The honest answer is no — there is no official, universal number that counts as a good Turnitin similarity score. Turnitin itself does not set a pass or fail threshold; it simply reports how much of your text matches sources in its databases. That said, a widely used rule of thumb at UK universities is that anything under roughly 15-20% is treated as an acceptable plagiarism percentage so long as the matched text is properly quoted, paraphrased and cited. The conditional half of that sentence matters far more than the number.
Different institutions, departments and even individual tutors apply their own expectations. A literature-heavy dissertation that quotes primary texts will naturally match more than a reflective essay written largely in your own words. So before you panic about a figure, the first question is not “what is a good similarity score?” but “what is my matched text actually made of?” If you want to understand the wider topic first, our explainer on what plagiarism is and the main types of plagiarism sets the groundwork.
What the similarity percentage actually means
This is the single most misunderstood point. The similarity score is a matching-text measurement, not a verdict on how much you plagiarised. A score of 18% means that about 18% of the words in your submission overlap with text Turnitin has seen before — in journal articles, websites, books, and previously submitted student papers. It does not mean 18% of your work is plagiarised.
Matched text and plagiarised text are not the same thing. Correctly quoted material, with quotation marks and a citation, is legitimate scholarship — yet it still shows up as a match. Your reference list will almost always match, because everyone cites the same sources in the same standard format. Common academic phrases (“the results suggest that…”, “a growing body of evidence”) match too. None of that is misconduct. Plagiarism is using someone else’s words or ideas without acknowledgement — and Turnitin cannot tell, on its own, whether a match is cited or stolen. Only a human reading the similarity report can make that judgement.
The score is context, not a verdict
Because the number is just matching text, two identical percentages can mean opposite things. A 25% score made up entirely of cited quotations and a standard methods section can be perfectly defensible. A 5% score that turns out to be one uncited paragraph copied word-for-word from a website is genuine plagiarism — and far more serious, despite the lower figure. The colour and the percentage are a starting point; the breakdown is the real story.
This is why experienced markers always open the report and read the matches. They look at where the matches fall (are they in your own argument, or in quoted blocks?), whether they are cited, and how long each matched string is. A scattering of 4-6 word matches across a paper is normal language overlap. A single unbroken 300-word match is a red flag regardless of the overall percentage.
“The similarity score is neither good nor bad on its own. Reviewing the Similarity Report is essential, since papers may contain matching text that is properly cited or quoted.” — Turnitin Guides, Interpreting the Similarity Report.
Understanding the Turnitin colour bands
Turnitin assigns a colour to each similarity report so you can gauge it at a glance. The bands are a visual shortcut, not a grading scale — but they are useful for triage. Broadly, the similarity index is shown as:
- Blue — no matching text (0%). Rare for real assignments with references.
- Green — one match to 24% similar. Usually the comfortable zone for most coursework.
- Yellow — 25% to 49% similar. Worth a careful look at the breakdown.
- Orange — 50% to 74% similar. High; expect scrutiny unless it is mostly legitimate quotation.
- Red — 75% to 100% similar. Very high and usually a serious problem.
Note that the colour bands describe the overall figure only. A green score can still hide a problematic uncited block, and an orange score can still be defensible if it is built from properly attributed quotations. Treat the colour as a prompt to investigate, never as a final answer.
Score range to interpretation: a quick reference
The table below summarises how similarity ranges are typically read at UK institutions. Treat it as guidance only — always check your own university’s academic-integrity policy and your assignment brief, which override any rule of thumb.
| Similarity score | Typical colour band | Usual interpretation |
|---|---|---|
| 0–5% | Blue / Green | Very low. Often just your name, headings, references and a few common phrases. Almost always fine. |
| 5–15% | Green | Low and typical for most essays. Generally accepted at many UK universities when matches are properly cited. |
| 15–24% | Yellow / Green | Acceptable in many cases, but worth reviewing. Check that matched text is quoted, paraphrased and referenced — not copied. |
| 25–49% | Amber / Orange | On the high side. Likely over-quoting or weak paraphrasing. Tutors will usually look closely; revise before submitting if you can. |
| 50–74% | Orange / Red | High. A large share of the text matches other sources. Strong risk of an academic-integrity query unless almost all of it is legitimately quoted material (rare). |
| 75–100% | Red | Very high. Usually signals heavy copying or a resubmission matching an earlier draft. Expect serious scrutiny. |
Two reminders before you read too much into the table. First, the acceptable percentage is often lower for short pieces: a 5% match on a 10,000-word dissertation is a few sentences, but 5% on a 500-word reflection could be a single copied paragraph. Second, resubmitting your own earlier draft can spike your score against yourself — ask your tutor to exclude prior submissions if that happens.
Why a “good” score depends on your assignment type
The same percentage carries different weight depending on what you are writing, because the nature of the task changes how much overlap is unavoidable. A “good” similarity score for a literature review is not the same as a good score for a creative reflection. It helps to think about why each type behaves the way it does:
- Literature reviews and dissertations quote and discuss many published sources, so they naturally run higher. A well-referenced literature review sitting in the low-to-mid twenties is common and usually fine, because most of the matches are cited engagement with the field rather than copying.
- Lab reports and methods sections reuse standard procedures, equipment names and fixed protocols, all of which match other students who ran the same experiment. A higher figure here is often structural, not suspicious.
- Law and case-study essays quote statutes, judgments and definitions verbatim — these cannot be paraphrased without changing their meaning, so they legitimately inflate the score.
- Reflective essays and creative pieces should be almost entirely your own words, so a low single-digit score is expected. A high figure here is a genuine warning sign.
The lesson is that you should compare your score against the norm for your kind of assignment, not against a single fixed number. If you are unsure what is normal for your module, your tutor or course handbook is the most reliable guide — and a draft check beforehand removes the guesswork entirely.
How Turnitin produces the number
Understanding the mechanics makes the percentage far less intimidating. When you submit, Turnitin breaks your text into overlapping strings and compares them against three broad collections: the open web (billions of crawled pages), a large repository of academic publications and journals, and a vast archive of previously submitted student papers. Where a string matches, it is highlighted and counted towards the similarity index. The score is simply the proportion of your submission that lights up.
Two consequences follow from this. First, Turnitin can only flag overlap with material it has already indexed — it is a similarity engine, not a plagiarism judge, which is exactly why a human still has to interpret the result. Second, the same passage can match a source you never read, because someone else used the same standard phrasing or cited the same authority. That is normal and is not evidence of wrongdoing. It is also why our own free plagiarism checker and the more detailed similarity report show you the matched passages themselves, so you can see whether each one is a harmless overlap or something that needs a citation.
What inflates a similarity score (often harmlessly)
A high-looking number is frequently caused by legitimate, well-cited content rather than copying. The usual culprits are:
- Quotations — every directly quoted sentence matches its source by definition. Heavy quoting inflates the figure even when each quote is perfectly cited.
- Your reference or bibliography list — standardised citation formats match almost word-for-word with everyone else who cited the same works.
- Common phrases and technical terms — stock academic phrasing and fixed terminology (legal definitions, standard methods, named tests) recur across thousands of papers.
- Cover sheets, headings and templates — required boilerplate and module titles match other students’ submissions.
- Block quotes and appendices — long extracts, survey questions or data tables you were asked to include.
The good news is that most of this is excludable. In the Turnitin Similarity Report you (or your tutor) can apply filters to exclude quotes, exclude the bibliography, and exclude small matches below a chosen word count. Doing so strips out the harmless overlap and shows the figure that actually reflects your own writing. If your tutor allows it, ask them to recalculate with quotes and references excluded before drawing any conclusion about the score.
How to legitimately lower your similarity score
Let’s be clear about what this section is and is not. There is no honest way to “beat” or “trick” Turnitin, and you should not try — character-swapping, hidden text and AI spinning are detectable and count as misconduct. Legitimately lowering your score means improving the originality of your writing so that less of it needs to match anything. Genuine methods include:
- Paraphrase properly, in your own words. Read, understand, close the source, then write the idea afresh — don’t just swap synonyms. Our guides on how to paraphrase and the best paraphrasing tools walk through this properly.
- Quote sparingly and only when the exact wording matters. Replace long block quotes with a short paraphrase plus a citation wherever you can.
- Cite everything correctly. Good referencing doesn’t lower the raw match, but it turns matches from misconduct into legitimate scholarship — see our guide to citing sources and the Harvard referencing reference for the standard formats.
- Add your own analysis. The more original argument, interpretation and synthesis you write, the smaller the proportion of matched text becomes.
- Exclude quotes and the bibliography in the report settings so the number reflects your own prose, not standard citations.
For a fuller step-by-step walkthrough of fixing flagged passages the right way, see our dedicated guide on how to remove plagiarism and reduce similarity. And remember that similarity is only half the integrity picture today — if your work involved generative tools, run it through our AI content detector as well, because many universities now check for AI-generated text alongside similarity.
Check your work free before you submit
See your likely similarity and spot uncited matches early with our free plagiarism checker (up to 3,000 words).
Check your likely score before you submit
Most students only see their Turnitin score after submission, when it is too late to fix anything. The smarter approach is to check your work in advance. You can run a draft through our free plagiarism checker to catch uncited matches and over-quoting while you still have time to revise. For a full Turnitin-level similarity report plus AI detection on a finished piece, our detailed plagiarism report gives you the same kind of breakdown your tutor will see.
Whatever the number comes back as, read the report, not just the percentage. Confirm every match is either your own wording, a properly cited paraphrase, or a correctly quoted and referenced passage. If it is, then even a higher-looking score is a good one — and that, far more than any magic threshold, is what “a good Turnitin similarity score” really means. If you are still unsure or working on a major project, our dissertation support and referencing resources can help you get it right before the deadline.
