Turnitin Similarity Report Explained: What the Percentage Means

A 0% on a Turnitin similarity report does not mean that your paper is clean. A 40% does not mean you cheated. Turnitin says so itself in their help pages. Most of the panic around this number comes from interpreting it as a verdict instead of as a pointer.

This post covers what the percentage is, what the colors mean, what instructors can exclude, and where the separate AI writing score fits in. Every figure below comes from Turnitin's published guides, which I read on 6 October 2026.

What the similarity score actually measures

The similarity score is the percentage of text that matches other sources. The score will range from 0 to 100. It is a count of matches and nothing else. Turnitin will highlight any passages of text that appear similar to other passages of text that the software has encountered previously. The user must determine whether these highlighted passages are problematic for the document.

Turnitin's guide for students explains that a high similarity score does not necessarily mean that a piece of writing was plagiarized, nor does a low similarity score show that no plagiarism occurred. A paper with long quotations that are properly cited will light up.

The color bands

The report colors the score so an instructor can scan a stack of papers quickly. The bands are the same in the two versions Turnitin documents, with one difference at the bottom of the scale. The colors can also vary depending on the way that your school has connected Turnitin to your learning platform, so the screenshot in a forum post may not be the same as your own screenshot.

Notice that the colors are not a grade. Green in one version is blue in the other. Red starts at 75%. The bands exist for sorting, and the guides do not attach a pass or fail to any of them.

Is there a good percentage?

No. According to Turnitin's student guide, there is no fixed number that must be received as a score on a Similarity Report and students should refer to their course syllabus or contact their instructor for further guidance. Any threshold that your school has established for the software is its own policy. Therefore, the number that matters to you is the one in your course documents.

A low score can also be a warning sign in the other direction. The student guide states that a low score may show that there is not enough research to support the writing. An essay that is meant to build on published work yet matches almost nothing may show that the author used few sources.

What can be excluded

Instructors can filter the report before they read it. The guide lists quotes and citations, bibliography entries, small matches and previous student submissions as things that can be excluded, and it suggests excluding matches of around 10 words or fewer so that a name or a stock phrase does not inflate the number. Excluding items recalculates the percentage, which is why the same paper can show two different scores depending on who opened it.

This is the practical reason to ask your instructor which view they use. If your bibliography is excluded on their side and included on yours, you are looking at different numbers for the same paper. Instructors also control whether students see the report at all. If yours is missing, one of the reasons is because the instructor turned off the student's access to the report.

The AI score is a different number

The second confusion is between the similarity score and the AI writing indicator. These are two separate measurements. Turnitin states that the percentage from its AI writing detection model is different from and independent of the similarity score, and that AI writing highlights are not visible in the Similarity Report.

The AI report has its own set of rules. A score of 0% shows that no qualifying text was identified as AI-generated. Scores between 1 and 19% are displayed as an asterisk instead of a numerical value, as Turnitin reports a higher incidence of false positives within this range of scores. From 20% up, the number represents qualifying prose that is likely to be AI-generated or AI-generated and then modified by a paraphrasing tool. The flagged text is highlighted in blue.

Turnitin states that the AI indicator should not be the sole reason for taking adverse action against a student and that human judgment and policy should be applied to the situation. If you are staring at an AI score you think is wrong, the longer treatment is in why an AI detector says your essay is AI when it isn't. For the mechanics of the model itself, see how Turnitin AI detection works.

How to keep a similarity score low for honest reasons

If your score is high due to quotes, the fix is not to remove them but to ensure that each quote is properly quoted, cited and short enough to do a job. If it is high because you leaned on a single source, the fix is to read more sources and say something of your own about them. A free citation generator helps with the formatting, but the underlying habit is the thing that moves the number.

If your concern is the AI indicator rather than the matching, that is a separate problem with a separate set of tools. The bypass Turnitin page covers that side, and the plain-English version is in does Turnitin detect AI.

The short version

The similarity score is a measure of how much of your text matches with other sources and is displayed as a percentage score between 0 and 100, with colors ranging from blue or green to red. This is not a plagiarism verdict, there is no official good number and instructors can exclude quotes, bibliography entries and small matches, which changes the figure. The AI writing percentage is a different measurement with its own asterisk rule below 20%. Ask your instructor which view they use and what threshold, if any, your course sets.

If you want the longer treatment:

• How Turnitin AI detection works — the mechanism behind the AI percentage
• Does Turnitin detect AI? — what it flags and what it misses
• What AI detector do colleges actually use? — which tools sit behind the dashboards
• Why an AI detector says your essay is AI — what to do when the score is wrong