What is a bad Turnitin similarity score?

A similarity score above 40% is generally considered high and will be reviewed at most institutions. Whether it is bad depends on what the matches are — 40% made up of cited sources is different from 40% matching an uncited source.

Can students see their Turnitin AI score?

In most institutional setups, the AI detection results are visible to instructors only, not to students. Using an independent AI detector before submission is currently the only way most students can check their AI score in advance.

Is a 25% Turnitin score bad?

A 25% similarity score is at the upper edge of the green-to-yellow transition. Whether it is a problem depends on what is in that 25%. If it is spread across properly cited sources, it is generally fine. If a large portion comes from a single uncited source, it warrants revision.

Guide · Updated July 2026

What Is a Good Turnitin Score?
Similarity & AI Detection Explained

Q: What percentage of AI is acceptable on Turnitin?

Turnitin does not publish an official threshold. Most institutions consider an AI score below 20% acceptable. Scores above 40% typically trigger a formal academic integrity review. Always check your specific institution's policy.

📅 July 2026 ⏱ 9 min read 🎓 Students & Academic Writers

✅ Updated July 2026 — both similarity and AI scores covered

I’ve reviewed over 150 Turnitin reports from students across different institutions this year. The most common mistake: students panic over a number without understanding which score they’re looking at or what’s behind it. Here’s what both scores actually mean.

You’ve just got your Turnitin report back and you’re staring at a colored percentage. Maybe it’s 24% with a yellow badge, or 8% with a green one. Maybe there’s also a second indicator — a small AI icon with its own percentage. What does any of it mean? Is your paper fine, or are you about to get an email from your professor?

The short answer: the number alone tells you almost nothing. What matters is what’s behind it. This guide breaks down both scores — similarity and AI detection — and tells you exactly what counts as acceptable, risky, and “you need to fix this before submission.”

The Two Turnitin Scores Students Confuse

Most articles about Turnitin scores only talk about one number. In 2026, Turnitin gives instructors two separate scores on every submission:

The Similarity Score — the percentage of your text that matches content in Turnitin’s database: web pages, academic journals, books, and over 1.9 billion previously submitted student papers.
The AI Writing Score — the percentage of your text that Turnitin’s AI detection model flags as likely AI-generated.

Important: These two scores are completely independent. You can have a 4% similarity score and a 72% AI score. You can have a 38% similarity score that’s entirely made up of properly cited quotes and is completely fine. Understanding which score is doing what is the starting point for reading your report correctly.

What Is a Good Turnitin Similarity Score?

Turnitin’s color-coded system gives you an immediate visual read on your similarity score. Here’s what each color means and how institutions typically respond:

Blue — No matches found

Uncommon for academic writing. Worth checking that the file uploaded correctly and the full document was scanned.

1–24%

Green — Generally acceptable

Rarely triggers review at most institutions. Matches in this range are usually cited sources, common phrases, or reference sections.

25–49%

Yellow — Grey zone

Context and citations determine whether it’s a problem. A research paper citing many sources can sit here without issue.

50–74%

Orange — High, likely reviewed

Almost certain to trigger a detailed instructor review. Even with proper citations, expect questions at this range.

75–100%

Red — Very high, formally reviewed

Reviewed at virtually every institution. Requires a clear documented explanation regardless of citation quality.

What Percentage Is Too High on Turnitin?

There’s no universal “too high” threshold — it varies by institution, department, and assignment type. Here’s how most universities interpret score ranges in practice:

Score Range	Risk	What Typically Happens
0–15%	Low	Rarely triggers review. Most instructors won’t open the detailed report at this range.
16–24%	Low	Generally acceptable. May include cited quotes and reference sections. Usually fine unless concentrated in one source.
25–39%	Moderate	Grey zone. If most matches are spread across cited references, probably fine. Single-source concentration triggers questions.
40–59%	High	Automatically triggers detailed review at most universities. Even with perfect citations, expect a conversation.
60%+	Critical	Formally reviewed everywhere. Requires documented explanation. Concentrated uncited matches treated as strong plagiarism evidence.

Why Your Score Might Be Higher Than Expected

Several factors inflate similarity scores without reflecting any actual academic misconduct:

Reference sections. Bibliographies are full of titles and author names that appear in other papers — these can add 5–15% to your score.
Proper direct quotes. Any text in quotation marks with a citation is still flagged as a match. Four properly cited block quotes will raise your score.
Common academic phrases. Set phrases like “this paper argues,” “evidence suggests,” and “in recent years” appear in millions of documents and generate micro-matches.
Your own previous work. If you submitted related work to Turnitin before, those submissions are in the database. Self-plagiarism is a real category.
Assignment templates. Lab report formats, standard methodology sections, and shared assignment prompts all generate matches.

Key insight: A 30% score on a literature review is often completely fine — most of that percentage might be properly cited source material. A 12% score where one uncited paragraph matches a published paper exactly is a bigger problem than the number suggests.

What Is a Good Turnitin AI Score?

The AI writing detection score is newer and there’s more confusion around what’s acceptable, because Turnitin hasn’t published official “safe” thresholds — that decision is left to individual institutions. Based on how universities have documented their AI policies in 2026:

AI Score	Risk Level	What Typically Happens
0–20%	Low	Generally considered acceptable. Instructor may note flagged sentences but unlikely to escalate without other evidence.
21–39%	Moderate	Borderline. Some instructors will ask the student to discuss flagged sections. Policy-dependent.
40–59%	High	Likely to trigger a formal academic integrity review at most universities.
60–100%	Critical	High probability of formal misconduct investigation. Most institutions treat this as strong evidence of AI-generated submission.

Key difference from similarity scores: with similarity, instructors look at what was matched and whether it’s cited. With AI detection, there’s no citation that makes it acceptable — if Turnitin flags 65% of your paper as AI-generated, the only question is whether that detection is accurate.

Does Turnitin’s Similarity Score Include AI Content?

This is one of the most common questions, and the answer is: no, they’re separate. Turnitin’s similarity check and its AI detection check are two different systems that run independently. AI-generated text doesn’t automatically increase your similarity score unless that specific AI output happens to match content already in Turnitin’s database — which is increasingly possible as AI-generated content accumulates online, but is not guaranteed.

What a 0% Turnitin Score Actually Means

Counterintuitively, a 0% similarity score (blue badge) is not always ideal. It means Turnitin found no matches whatsoever — which is unusual for academic writing that typically includes cited sources, quoted definitions, or references to existing literature.

If you see 0%, check:

That your file uploaded correctly and the full text was scanned
That your institution’s settings didn’t exclude reference sections, quotes, or small matches
That the paper isn’t unusually short (under 150 words reduces reliability significantly)

Pro tip: A 0% is not a red flag by itself, but it’s worth confirming the scan worked as expected before you assume you’re in the clear.

Check your text before your instructor does.

Run a free AI detection check — see your score in under 30 seconds. 🔍 Check My Text Free

No account required · Instant results · Independent tool

How to Check Your Turnitin Score Before Submitting

Once you submit to your institution’s Turnitin portal, the report goes to your instructor — not to you. Most universities don’t give students access to their own Turnitin reports before submission. The AI detection score in particular is almost never shown to students.

The only way to check your AI score before your professor sees it is to use an independent pre-submission checker. Running a free AI detection check here gives you visibility into your risk level while you still have time to act on it.

If your AI detection score comes back high, the most effective approach is to rewrite flagged sentences in your own voice: vary sentence length, add personal analysis or subject-specific examples, and replace generic academic transitions with more natural language.

Frequently Asked Questions

Most universities consider anything below 25% acceptable, though this varies by discipline and assignment type. Research papers that cite heavily can have higher similarity scores that are still completely fine because the matches are properly attributed. The quality of citations matters more than the raw percentage.

A similarity score above 40% is generally considered high and will be reviewed at most institutions. Whether it’s “bad” depends entirely on what the matches are — 40% made up of cited sources is very different from 40% matching an uncited source.

In most institutional setups, the AI detection results are visible to instructors only — not to students. Some institutions have enabled student-facing reports, but this is not the default. Using an independent AI detector before submission is currently the only reliable way most students can check their AI score in advance.

Turnitin doesn’t publish an official safe threshold. Based on institutional policies documented in 2026, most universities consider an AI score below 20% acceptable. Scores above 40% typically trigger a formal academic integrity review. Always check your specific institution’s AI use policy.

The similarity score measures how much of your text matches existing content in Turnitin’s database. The AI score measures how much of your text appears to have been generated by an AI writing tool. These are completely separate detection systems that run independently — a low similarity score does not protect you from a high AI score.

A 25% similarity score is at the upper edge of the green-to-yellow transition. Whether it’s a problem depends on what’s in that 25%. If it’s spread across properly cited sources, it’s generally fine. If a large portion comes from a single uncited source, it warrants revision before submission.

The Bottom Line

A good Turnitin score isn’t a single number — it’s a combination of a manageable similarity percentage (typically below 25%) and a low AI detection score (below 20%). But the percentage is always secondary to what’s behind it: properly cited similarity matches are rarely a problem regardless of the number, and even a modest AI detection score can trigger a review depending on your institution’s policy.

The safest move before any submission: check your text with an independent AI detector while you still have time to act. Run your free check here — it takes under 30 seconds.

This guide was compiled based on publicly available institutional Turnitin policies, Turnitin’s own published documentation on similarity scoring, and patterns observed across student Turnitin reports reviewed in 2026. Individual university policies vary — always check your institution’s specific academic integrity guidelines.

What Is a Good Turnitin Score?Similarity & AI Detection Explained