What Is a Turnitin Similarity Score? How to Read It
Few things cause as much anxiety as seeing a similarity percentage next to your paper. Students worry about numbers like 18 percent or 31 percent, and rumors circulate about "safe" limits. The reality is more nuanced, and understanding what the score actually shows can save you a lot of stress.
This guide explains what a Turnitin similarity score is, how the report works, what instructors look at and what you can do to use the tool sensibly. Institutions configure the software differently, so treat this as general guidance and follow your own course instructions.
What the score is
Turnitin is a text-matching service used by many schools and colleges. When you submit a paper, the software compares it with a large collection of material: web content, published articles and books, and papers that have been submitted to the service. The similarity score is the percentage of your paper that matches text found in those sources.
The important word is similarity. The tool finds overlap. It does not decide whether the overlap is acceptable, and it does not decide whether anyone cheated. That judgment belongs to the instructor.
What the similarity report shows
The report is more detailed than the number.
- The percentage of matched text overall.
- Highlighted passages in your paper, color-coded by source.
- A list of matched sources, with the share each one contributes.
- Links to the matched material, so your instructor can compare the passages.
- Filters that can exclude quotations, the bibliography or very small matches, depending on the settings.
Instructors usually read the highlighted passages, not just the number. A report that is 25 percent made up of properly cited quotations and a long reference list tells a very different story from one that is 25 percent copied paragraphs.
Why there is no universal safe number
Students often ask what percentage is acceptable. There is no official answer, and any number you hear is a rumor or one instructor's habit. Several things affect what a score means.
- The type of assignment. A lab report with a standard template or a paper with many quotations may naturally score higher.
- The length of the paper. Short papers can produce large percentages from a few matched sentences.
- The settings. Whether quotations and bibliographies are excluded changes the number.
- The sources. Matches to a single source can be more worrying than many small matches across many sources.
- The context. A shared assignment prompt copied into the paper can create matches.
Because of this, a high score can be innocent, and a low score can still hide a problem.
Matches that are usually harmless
Some overlap is normal and expected.
- Quotations that are properly marked and cited.
- The reference list, especially when many students cite the same sources.
- Standard phrases, such as the names of institutions, titles or technical terms.
- Assignment questions or headings copied from the prompt.
- Common formulas, definitions and templates where wording is fixed.
If these are filtered or understood in context, they should not cause concern.
Matches that raise questions
Instructors look more closely at:
- Long passages matching a single source with no quotation marks or citation.
- Close paraphrases that follow a source's structure.
- Matches to another student's paper, which may indicate collusion.
- Large blocks of text from websites, especially essay-sharing sites.
- Sources cited in the paper that do not appear to support what it says.
If you are called to discuss a report, the best response is to look at the specific passages and explain them honestly.
How instructors typically read a report
While practices vary, instructors usually:
- Look at the overall percentage as a prompt for attention, not a verdict.
- Open the highlighted passages and the matched sources.
- Check whether matches are quoted and cited.
- Consider the assignment type and the student's usual style.
- Decide whether the matches show poor citation habits, carelessness or deliberate copying.
- Decide whether to give feedback, ask for a rewrite or start a formal process.
For many low-level issues, the outcome is feedback about how to cite, not a penalty.
AI writing indicators
Some institutions also use tools that estimate whether text was written by AI. These are different from similarity checks, and their reliability is debated. Detection results are generally treated as one piece of information, not proof, and many institutions caution against relying on them alone. If your course has rules about AI, follow them, and keep drafts or notes that show how your work developed. Our guide to citing AI tools covers how to acknowledge permitted use.
Using a draft checker sensibly
If your course offers a draft check, use it as a learning tool.
- Run it early, so you have time to fix problems.
- Read the highlighted passages, not just the percentage.
- Fix real problems: add citations, rewrite close paraphrases, mark quotations.
- Do not chase a number. Removing a properly cited quotation to lower the score makes the paper worse, and it does not make it more honest.
- Do not rewrite by swapping words to fool the software. That is patchwriting, and a human reader may still see it.
What to do if your score looks high
- Look at what is matching. Often it is quotations or the reference list.
- Check your citations on any matched passage you did not quote.
- Rewrite genuinely copied passages in your own words with a citation.
- Ask your instructor if you are unsure how the report is read or whether you can resubmit.
- If you are asked to explain, be calm and specific, and bring your notes and drafts.
If you realize that you copied something by mistake, it is usually better to say so early.
Myths about similarity checking
- "Under 15 percent is safe." There is no such rule, and instructors differ.
- "Zero percent is best." A paper with no matches to common phrases, quotations or references is unusual, and it does not prove originality.
- "Changing a few words beats the checker." It does not beat a human reader and it is still plagiarism.
- "Turnitin catches everything." It only compares with what is in its database.
- "Turnitin decides if I cheated." People decide, using the report as one piece of evidence.
- "My paper is stored forever and anyone can read it." What happens to submitted papers depends on your institution's settings and policies, so check them.
Common mistakes
- Panicking about the number instead of reading the matches.
- Removing citations to reduce the score.
- Submitting early drafts you have not checked.
- Ignoring the institution's policy on how reports are used.
- Assuming a low score is proof of originality.
For the wider rules, see our guide to avoiding plagiarism.
A worked example of reading a report
Imagine a 2,000 word essay with a 22 percent score. On opening the report, the instructor sees that about 8 percent comes from the reference list, 7 percent from three properly quoted and cited passages, and 4 percent from short, common phrases such as the name of a government agency. The remaining 3 percent is one paragraph that matches a web page closely, with a citation at the end but no quotation marks. The headline number looks alarming, but the real issue is one paragraph that needs rewriting or quoting.
Now imagine a second essay with a score of 6 percent where the whole of the matching text is a single uncited, copied page from an essay-sharing site. The number is lower, but the problem is far more serious. This is why instructors read the passages and not just the percentage.
A quick summary
- The score shows overlap, not misconduct.
- No percentage is automatically safe or unsafe.
- Instructors read the highlighted passages in context.
- Quotations, references and standard phrases are normal matches.
- The best protection is honest citation, careful paraphrase and early checking.
Final thoughts
A similarity score is a screening tool, not a judge. If you quote properly, paraphrase honestly and cite as you write, you have little to fear from the report. Use any draft checker to learn, read the highlighted matches instead of the percentage and keep your notes so you can explain how your work developed.
Frequently asked questions
What is a Turnitin similarity score?
It is the percentage of text in your paper that matches text in sources in Turnitin's database, such as web pages, published works and previously submitted papers. It shows overlap, not whether plagiarism occurred.
What is an acceptable similarity score?
There is no universal number. Instructors and institutions decide how to interpret the report, and a low score can still hide a problem while a higher score can be entirely innocent. What matters is what the matches are.
Does a high score mean I plagiarized?
Not necessarily. Properly quoted and cited passages, reference lists, common phrases and assignment prompts can all produce matches. A human reader has to look at the highlighted passages.
Does a low score prove my work is original?
No. A low score only means the text did not match much in the database. It cannot detect everything, such as work from sources not in the database, translated text or ideas that were paraphrased without credit.
Can I see my own report?
It depends on your instructor and institution. Some give students a draft checker or share the report, and some do not. Ask your instructor or check your course page.