There is no universal “acceptable” Turnitin similarity score. The number you see is a measure of matching text against Turnitin’s database, not a verdict on plagiarism, and most Australian markers treat anything roughly under 15 to 20 percent as unremarkable once quotes and references are accounted for. What matters is whether the matched text is properly cited and paraphrased, not the percentage by itself.
If your tutor has asked you to keep your similarity down, the legitimate path is better referencing and genuine paraphrasing, not tricks to hide copied work. This guide explains what the score actually counts, why no single percentage is “safe”, how to read the typical bands, and how to lower your score the right way. If you want a second pair of eyes before you submit, our proofreading and editing service can flag weak paraphrasing and missing citations without changing your argument.
Key takeaways
- It measures matches, not cheating: the similarity score shows how much of your text matches other sources, including your own properly quoted references.
- No magic number: there is no official “acceptable” percentage; the threshold depends on your unit, the assessment type, and your marker’s judgement.
- Quotes and references inflate it: long quotations, reference lists, and common academic phrases can push the figure up even when your work is original.
- Reduce it legitimately: paraphrase in your own words, cite and quote correctly, add your own analysis, and exclude the bibliography if your uni allows.
- Very low can be a flag too: a near-zero score sometimes signals missing citations or a file Turnitin could not read properly.
What the Turnitin similarity score actually measures
The Turnitin similarity score measures the proportion of your submission that matches text in Turnitin’s database. That database includes published journals, books, websites, and millions of previously submitted student papers. When Turnitin finds a string of words that lines up with a source, it highlights it and adds it to the percentage. The Similarity Report is a colour-coded map of those matches, not an accusation.
This distinction matters because a high number does not automatically mean plagiarism, and a low number does not automatically mean integrity. A literature review that quotes several definitions verbatim, with perfect citations, can score 25 percent and be completely honest. A paraphrased essay that lifts ideas without acknowledgement can score 6 percent and still breach your university’s academic integrity policy. The percentage is a starting point for a human reviewer, not the conclusion.
Turnitin also cannot, on its own, judge intent or whether content was generated by a chatbot. If you are worried about that side of things, our companion piece on whether Turnitin detects AI writing covers how the AI indicator works and why it is separate from the similarity score. For a broader look at the same question, our guide on whether AI can be detected in an assignment explains what these tools actually flag, and our walkthrough on how to check if your assignment reads as AI-written shows how to self-review before you submit.
Why there is no single “acceptable” percentage
Students often ask for one safe number, and tutors rarely give one, because the meaning of any figure changes with context. A 12 percent score on a 1,500 word reflective essay, where most of the text should be your own voice, reads very differently from 12 percent on a heavily referenced clinical case study at a Group of Eight uni. The expectation also shifts with how the piece is built, so it helps to understand the way a clear university essay structure distributes original argument against cited evidence. Australian universities such as the University of Sydney, Monash, and UNSW set their own guidance, and individual unit coordinators frequently add their own expectations on top.
Several legitimate things push your score up without any wrongdoing on your part. Direct quotations, which you are allowed and sometimes required to use, count as matches. Your reference list matches every other paper that cited the same sources. Standard academic phrasing (“the results suggest that”, “a significant body of literature”) matches thousands of documents. Assignment questions and templates pasted into your file match every classmate who used the same brief. None of this is plagiarism, yet all of it inflates the number. If you have used a writing assistant to brainstorm or draft, it is worth knowing the boundaries first, which our guide on whether you can use AI for university assignments sets out in line with current Australian integrity expectations.
How to read the score bands (with caveats)
The table below gives a rough interpretation of common similarity bands. Treat it as a guide for self-review only. Always defer to the exact threshold in your unit outline or your marker’s instructions, because these override any general rule of thumb.
| Similarity band | Typical interpretation | What to check |
|---|---|---|
| Under 10% | Usually low concern, common for original work with light quoting. | Confirm citations are actually present; a near-zero score can mean missing references. |
| 10% to 20% | Common and often fine, especially with a reference list and some quotes. | Open the report; make sure matches are quotes, references, or common phrasing, not copied paragraphs. |
| 20% to 35% | Worth a closer look; may be acceptable for heavily referenced work. | Check for one large single-source match and paraphrase or quote it properly. |
| Over 35% | Investigate before submitting; often signals weak paraphrasing. | Rewrite matched passages in your own words and confirm every source is acknowledged. |
For example, a UTS student who submits a referenced report and sees 22 percent may find that almost all of it comes from their bibliography and three properly cited quotes. After excluding the reference list in the report settings, the figure could drop to single digits with no rewriting at all. The percentage moved, but the integrity of the work never changed. The same logic applies to AI indicators, where panic over a flag is rarely justified; our explainer comparing an AI-written assignment against human-written work shows why context, not a raw figure, decides the outcome.
How to reduce your similarity score legitimately
The honest goal is not a lower number for its own sake. It is writing that genuinely reflects your own understanding, with every borrowed idea acknowledged. When you do that well, the score falls as a side effect. Here is the legitimate process.
- Paraphrase properly, not word-swap. Read the source, close it, and write the idea in your own sentence structure from memory, then cite it. Replacing a few words while keeping the original sentence shape still matches and still risks an integrity breach. This genuine rewriting is also what separates honest work from the shortcuts described in our guide on writing an assignment without leaning on AI detection tricks.
- Quote correctly and sparingly. Use quotation marks for any text taken verbatim, keep quotes short, and add the in-text citation. Correct quoting is honest, even though it adds to the percentage.
- Lead with your own analysis. Markers reward your interpretation, comparison, and critique. The more original argument you add between sources, the lower the proportion of matched text becomes naturally.
- Reference accurately. Follow your required style precisely, whether that is APA, Harvard, or AGLC. A clean reference list is expected; our Harvard referencing guide, APA 7 referencing guide, and AGLC4 referencing guide walk through the formatting. If any of your sources are AI tools, our note on how to reference ChatGPT shows the correct way to acknowledge them.
- Exclude the bibliography and quotes if allowed. Many Turnitin settings let you exclude the reference list and quoted material. Check whether your unit permits this before assuming the raw number is the final one.
Aim to lower your similarity by understanding your sources better, never by disguising copied text.
BAO academic support team
Not sure your paraphrasing will hold up? Our Australian editors check your draft for weak paraphrasing, missing citations, and quote formatting, returning AI-free, Turnitin-checked feedback matched to your rubric.
Why a very low score can also be a flag
A score close to zero is not always a good sign. If your assignment cites multiple sources, a literature review with almost no matches can suggest the citations are missing, mislabelled, or that Turnitin could not read your file. Image-based PDFs, locked documents, or text saved as pictures sometimes return artificially low figures because the software simply could not scan the words.
It can also raise questions for an experienced marker. A research essay that quotes definitions and references a dozen studies would normally show at least some matching text. A flat zero on that kind of task can prompt a tutor to look more closely, not less. The lesson is the same in both directions: chase honest, well-cited writing rather than a particular number on the dial. A literature review is a good example, since it is built almost entirely from sources; our walkthrough on how to write a literature review shows where citations should naturally appear.
How Australian universities use the score alongside judgement
Australian universities use the similarity score as a prompt for human review, never as an automatic penalty. A marker or academic integrity officer opens the full Similarity Report, looks at where the matches sit, and decides whether they reflect honest quoting and referencing or something that needs investigating. Policies at institutions like RMIT, the University of Melbourne, and Macquarie all frame Turnitin as a teaching and checking tool, with the academic making the final call.
This is why panicking over a single percentage is rarely useful. A 30 percent score made up of your reference list and correctly cited quotes is defensible. A 9 percent score where one uncited paragraph was lifted from a website is not. If you want a clear-eyed look at what actually triggers an academic integrity case, our explainer on whether you will get caught sets out how detection and review really work in practice.
The practical takeaway for any Australian student is to read your unit’s integrity guidance first, open your Similarity Report rather than trusting the headline number, and fix the substance. Strong paraphrasing, accurate citations, and your own analysis will keep both your marker and the software satisfied.
Frequently asked questions
What Turnitin similarity score is acceptable?
There is no single acceptable Turnitin similarity score, because the threshold depends on your unit, the assessment type, and your marker’s judgement. As a rough guide, many Australian markers treat scores under 15 to 20 percent as unremarkable once quotes and references are accounted for. Always check your unit outline, since some assessments expect lower and some heavily referenced tasks tolerate more.
Does the Turnitin similarity score mean I plagiarised?
The Turnitin similarity score does not mean you plagiarised; it only measures how much of your text matches other sources. Properly cited quotes, your reference list, and common academic phrases all add to the figure without any wrongdoing. A human marker reviews the actual matches to decide whether there is an integrity issue.
Why is my Turnitin similarity score so high when my work is original?
Your Turnitin similarity score can be high even when your work is original because quotations, reference lists, assignment templates, and standard academic phrasing all count as matches. A heavily referenced report naturally shows more matching text than a reflective essay. Open the report to confirm the matches are legitimate quotes and references rather than copied paragraphs.
How can I reduce my Turnitin similarity score legitimately?
You can reduce your Turnitin similarity score legitimately by paraphrasing ideas in your own words, quoting and citing correctly, adding more of your own analysis, and excluding the bibliography and quotes if your uni allows. Avoid word-swapping tools or hidden characters, since these breach academic integrity and can be detected. The aim is genuine understanding, not a disguised copy.
Is a very low Turnitin similarity score a problem?
A very low Turnitin similarity score can be a problem if your assignment cites sources but shows almost no matches, which may signal missing citations or a file the software could not read. Image-based PDFs and locked documents sometimes return artificially low figures. Check that your references appear and that your file is readable text before assuming a low score is good news.
Can I exclude quotes and references from my Turnitin score?
You can often exclude quotes and references from your Turnitin score, as the report settings allow the reference list and quoted material to be removed from the calculation. This usually drops the percentage without changing a word of your work. Confirm your unit permits these exclusions before relying on the adjusted figure.
Do Australian universities fail you for a high similarity score?
Australian universities do not fail you automatically for a high similarity score, because the number is a prompt for human review rather than a penalty. A marker examines where the matches sit and whether they reflect honest quoting and referencing. A high score built from citations is defensible, while a low score hiding an uncited passage is not.
Is the Turnitin similarity score the same as the AI score?
The Turnitin similarity score is not the same as the AI writing indicator; they are separate measures. The similarity score reports matching text against existing sources, while the AI indicator estimates whether content was machine-generated. The two reports are read independently, so a clean similarity score says nothing about the AI indicator and vice versa.