Turnitin score meaning
A Turnitin score can mean two different things: the classic similarity score and the newer AI-writing score. Here is what each one measures, how schools usually interpret them, and when you should worry.There are really two Turnitin scores
The classic Turnitin number is the similarity score. It estimates how much of the document overlaps with sources in Turnitin's corpus. The newer number, when enabled by the institution, is the AI-writing score. That score estimates how much of the document resembles model-generated writing.If you are looking for the broader context, read the full Turnitin guide. If you only need the detector-specific version, go to does Turnitin detect AI.What the similarity score means
Similarity is not plagiarism. It is overlap. A bibliography, direct quotations, template language, and assignment prompts can all increase similarity without proving misconduct. That is why one instructor may be comfortable with a percentage that another instructor wants to inspect.The right question is not "is 18% good?" in the abstract. The right question is whether the matched text is expected, cited correctly, and consistent with the rules of the assignment.What the AI score means
The AI score is different. It is not checking source overlap at all. It is estimating how much of the prose behaves like LLM output. Most schools treat that number as a screening signal, not as an automatic verdict.Turnitin itself has repeatedly framed the detector as an input into human review, especially for lower-percentage results. If you want the research angle on why those edge cases matter, see our page on Turnitin false positives.A practical way to read the number
- Low similarity, low AI: usually the cleanest case.
- High similarity, low AI: likely a citation or overlap issue, not an AI issue.
- Low similarity, high AI: classic AI-detection problem; the prose pattern is the concern.
- High similarity, high AI: the draft may have both source-overlap and detector-risk problems.
Check your text before you submit
If you are worried about how your draft will read to a detector, you can run it through Leap's free AI detector in your browser. There is no signup and nothing is sent anywhere. It scores your text from writing signals like sentence-length variation, stock AI phrases, hedging, and repetition, and highlights the sentences that weigh most. Treat it as one signal, not proof.What students should do before submitting
- Check whether the number you are seeing is similarity, AI, or both.
- Review matched passages if it is a similarity issue.
- Use a consumer detector before submission if AI risk is the concern.
- Keep version history and notes in case you need to defend original work later.
Frequently asked questions
No. Most of the time, 'Turnitin score' refers to the similarity score. Some instructors also see an AI-writing score. They are separate measurements.
There is no single universal threshold. A low score is generally easier to defend, but acceptable overlap depends on assignment type, quotations, citations, and instructor policy.
Many institutions treat scores above roughly 20% as a reason to review the submission more closely, not as automatic proof of misconduct.
Yes. Similarity is about overlap with sources. AI detection is about the writing pattern itself. You can score low on one and high on the other.
Keep your draft history, check the paper with another detector, and be ready to explain your writing process. A single Turnitin score should not be the whole case against you.