A Turnitin similarity score comes back as a number and a color, and it's easy to assume the color is a verdict — green means fine, red means caught. Turnitin's own help documentation says otherwise: the score measures matching text, not plagiarism, and it explicitly leaves the "what's acceptable" question to your institution. Here's what the bands actually mean, and what real university policies set as their cutoff.
Trusted by students at
Turnitin's own help documentation defines five similarity bands by color: blue for 0% matching text, green for one word up to 24%, yellow for 25–49%, orange for 50–74%, and red for 75–100%. The icon color is simply a visual read on "the amount of matching or similar text that was uncovered" — that's the full extent of what Turnitin's own documentation says the color represents.
Notably, Turnitin's documentation doesn't say anywhere that a given band means plagiarism occurred or didn't. A paper can land in the yellow or orange band entirely because of properly quoted material, a long reference list, or a shared assignment template — all of which show up as "matching text" whether or not anything improper happened. Turnitin builds the detector; your school decides what the number means.
Because Turnitin doesn't define a pass/fail line, individual universities publish their own. Westcliff University's similarity policy for dissertations states plainly: "Generally, 15% will be considered the maximum acceptable level on the Similarity Report" — but it adds that a dissertation can still be rejected below that number if continuous unquoted blocks of text are flagged.
The University of Ghana's Turnitin usage guidelines set a different structure entirely: an overall similarity index that "should not exceed 20%," a single-source match that should stay under 2%, and no unbroken matching string longer than 10 words — with all three conditions needing to hold, not just the headline percentage. Both are real, published policies, and they don't agree with each other on the number. That's the actual answer to "what's a good score": whatever your own syllabus, handbook, or supervisor's stated threshold says, because there isn't one industry-standard figure.
The matches that make up a similarity score are often the unremarkable parts of a paper: direct quotes with proper citations, a bibliography formatted to a required style, a shared lab report template, or common phrasing in a field with limited ways to say something precisely. None of that is plagiarism, but all of it registers as "matching text" the same way copied prose would.
StudyPilot's plagiarism checker gives you a similarity percentage and a source-by-source breakdown before you submit anything, scanning against web pages, academic papers, and a database of previously submitted student work — similar in structure to what Turnitin checks, though it's a separate index, so the exact number won't match your school's Turnitin report one-to-one. It's built as a pre-submission sanity check, not a prediction of your official score, and your essay is scanned and then discarded rather than stored. If something gets flagged, you can rewrite the matched sentence in one click rather than guessing at a fix.
Paste your essay into StudyPilot's plagiarism checker for a source-by-source originality report in under a minute — then fix any flagged sentence with a one-click rewrite.
Check My Essay Free →