What Is Turnitin Similarity Score?
A Turnitin similarity score is the percentage of a submitted paper that matches text in Turnitin's index of web pages, publications and prior student submissions. It measures matching, not cheating: quoted, cited and common phrases all raise it, and instructors read the report, not just the number. This page unpacks the definition, the mistakes graders see most, and a worked example you can pattern-match against your own draft.
James Parker, MLS
Citation & Formatting Reviewer · Master of Library Science; 9 years as a reference librarian coaching APA, MLA and Chicago styles
Quick answer
A Turnitin similarity score is the percentage of a submitted paper that matches text in Turnitin's index of web pages, publications and prior student submissions. It measures matching, not cheating: quoted, cited and common phrases all raise it, and instructors read the report, not just the number.
On this page · 4 sections
The definition, precisely
When a paper uploads through the LMS, Turnitin breaks it into overlapping word patterns and compares them against three indexes: the live web, a library of publications, and a repository of every student paper ever submitted through the service. The similarity score is the percentage of your text that matched something — nothing more. Alongside it sits the report itself: a marked-up copy where each matched passage is color-coded to its source, which is the document instructors actually read. The report is the evidence; the percentage is only its summary line.
Interpreting the number is contextual, and Turnitin publishes no official threshold. Rough conventions: under 15 percent is typical for properly cited work, 15-30 percent invites a look at what matched — often the quoted passages and the reference list itself — and above 30 percent usually means quotes or boilerplate need trimming. Every legit match is visible in the report: the block quote you cited, the assignment prompt your instructor requires, your own reference list. A score is a conversation starter between a draft and its reader, never a verdict.
Lowering the score legitimately and lowering it illegitimately look identical to the number and opposite to a reader. Legitimate: paraphrase with citation instead of over-quoting, analysis in your own voice, quotes reserved for language worth quoting. Illegitimate: swapping synonyms until sentences stop meaning anything, inserting hidden characters, or running paraphrase tools that mangle prose — instructors who read the flagged passages spot word-salad instantly, and mangled paraphrase is still patchwriting when the structure survives. The integrity judgment happens in the read, not the percentage.
The four mistakes graders see most
Every one of these shows up in a stack of sixty papers — usually more than once. Each pairs the error with the fix, so the table doubles as a pre-submission check.
| Common mistake | The fix |
|---|---|
| Treating the similarity score as a plagiarism verdict. | It is a match report. Quotes, references and common phrases raise it honestly; the instructor reads the sources behind the percentage before judging anything. |
| Panicking at 25 percent without opening the report. | Check what matched. Six points of cited block quote plus four of reference list plus boilerplate headings reads clean — the breakdown is the information. |
| Running text through a synonym tool until it stops making sense. | Mangled paraphrase is detectable by any human reading it, and patchwriting still violates codes when the sentence structure survives. Paraphrase by rebuilding the idea, not by swapping words. |
| Assuming a low score proves originality. | A 3 percent paper can still be uncited plagiarism of an unpublished source the index lacks. Originality lives in the citation practice, not the percentage. |
A worked example
A student's 2,000-word paper returns 18 percent. The report shows the composition: 6 percent is a properly quoted and cited passage from a journal article, 4 percent is the reference list matching itself, 8 percent is short overlapping phrases from cited sources — standard disciplinary vocabulary. Her instructor reads the breakdown, confirms every match traces to a citation, and moves on to grading the argument. The 18 was never the finding; the read was. Her roommate's 6 percent, built on uncited paraphrase of an unindexed blog, was the real problem.
Pattern-match your own draft against this example rather than against the definition: concrete cases transfer better than abstract rules, and the example shows what the rule looks like when it is done right.
How turnitin similarity score differs from its neighbors
A similarity score is not an AI-writing detector: Turnitin's AI-detection product is a separate output with different logic — pattern-based, probabilistic, and treated with caution by most institutions — while similarity is a literal text-matching percentage. It is not a plagiarism judgment, which is a human decision about attribution and intent. And it is not comparable across assignments: a lab report with a required methods boilerplate scores higher than a personal essay by design, so the number only means anything beside its own breakdown.
Boundary confusion is where most grading deductions live: the term is understood roughly but applied against the wrong neighbor. Pin the boundary above and the term stops being fuzzy in practice, which is what the rubric is actually testing.
What's your next step?
Two taps to the exact page that unblocks this paper.
The 30-second check
Skim the four mistakes above against your draft. Any hit is a fix you can make in minutes — and the format row of the rubric is where they all live.
Run it through the tool
The free synonymizer handles the mechanical half of this term — check it before you submit.
Open the toolRelated term: Rubric — the neighbor this term is most often confused with. Reading the pair together is how the boundary sticks.
Frequently asked questions
4 answered1What similarity score is too high?
There is no universal threshold; instructors read reports, not just percentages. As rough context: under 15 percent is typical for cited work, 15-30 percent gets a look at the matches, and over 30 usually means quotes, boilerplate or reference-list text needs trimming. The breakdown matters more than the headline number.
2Can I check my paper before submitting it?
Some institutions let students self-submit a draft to a practice class; some writing centers run checks on request. Ask your instructor first — a self-check against the same repository also stores your draft, which some courses handle differently. Never trust a third-party scanner with your paper before submission.
3Why did my correctly cited quotes raise the score?
Because Turnitin matches text, not integrity — a quotation mark does not change the words. Most instructors exclude quotes and the bibliography from the effective score, or simply expect them there. If your quotes are legitimate, the flagged color on the report is documentation, not accusation.
4Does paraphrasing lower the score?
Honest paraphrase does: rebuilding the idea in your own sentence structure with a citation produces little matching text. Synonym-swapping does not — the structure still matches, and the prose reads wrong. If a tool did the paraphrase, assume the flagged passage will be read closely.
James Parker, MLS
Citation & Formatting Reviewer · Master of Library Science; 9 years as a reference librarian coaching APA, MLA and Chicago styles