Turnitin Score Meaning: Percentage, Color & More Explained
What does your Turnitin score actually mean? Here's how to read the similarity percentage, understand the color codes, and interpret AI detection results.
You've submitted your paper, and now you're staring at a number with a colored badge next to it. Is 24% good? Is 5% suspicious? What does that blue icon on the AI tab mean?
This guide breaks down everything about Turnitin scores — the similarity percentage, the color system, the AI detection indicators, and what actually counts as a "good" score.
What Does the Similarity Report Actually Detect?
Turnitin doesn't detect plagiarism directly — it identifies text that matches other sources. The platform searches billions of web pages (live and archived), academic journals, books, and previously submitted student papers. When it finds matching content, it flags it for instructor review.
The similarity score is simply the percentage of your paper that matches content in Turnitin's database. It doesn't tell you whether those matches are properly cited quotations, common phrases, or actual plagiarism — that's for your instructor to determine.
What the Color of the Similarity Score Tells You
Turnitin uses a color-coded system to give you a quick visual read on your similarity level:
| Color | Similarity Range | What It Means |
|---|---|---|
| Blue | 0% | No matching text found in the database |
| Green | 1–24% | Low similarity — generally acceptable for most assignments |
| Yellow | 25–49% | Moderate similarity — worth reviewing the matched sections |
| Orange | 50–74% | High similarity — significant portions match other sources |
| Red | 75–100% | Very high similarity — most of the paper matches existing sources |
Exploring the Match Overview Panel
When you open a similarity report, you'll see several sections in the panel:
- Match Overview — the overall similarity percentage with highlighted matching sections in your paper
- All Sources — a list of every flagged source (journals, websites, student papers) with the percentage each one contributes
- Filters and Settings — options to exclude quotes, bibliographies, or small matches to get a clearer picture
- Excluded Sources — any sources that have been manually removed from the comparison
- Flags Panel — detects potential cheating attempts like invisible text, hidden characters, or symbol replacement
Real Examples: What Different Scores Can Mean
Example 1: Score inflated by student headers
A student's name, course number, and professor name appear on every paper in the class. Turnitin flags these as matching text unless your instructor enables the "exclude small matches" filter.
Example 2: 100% match from resubmission
A student submits a draft, then resubmits the final version. The second submission shows 100% similarity because it matches the first submission already stored in Turnitin's database.
Example 3: Two papers at 20% — very different situations
Paper A has 20% similarity from properly cited quotations and references — perfectly fine. Paper B has 20% similarity from two paragraphs copied word-for-word without attribution — a potential integrity issue. The number is the same; the context is completely different.
Example 4: 53% drops to 12% after filtering
A research paper heavy on quoted material shows 53% similarity. After the instructor excludes direct quotations and bibliography entries, the score drops to 12% — revealing that the student's original analysis is solid.
How Turnitin Detects Student Collusion
Turnitin scans all submissions again after the assignment due date. This means if Student A submits first and Student B copies their work, Student A's paper may show a low similarity score initially, while Student B's will show a very high match against Student A's stored submission.
This is why some instructors configure assignments to add papers to Turnitin's repository — it creates a record that makes collusion much easier to detect.
Understanding Turnitin's AI Detection Score
In addition to the similarity score, Turnitin now provides an AI detection score that estimates what percentage of your paper was likely written by AI tools like ChatGPT. This score is only visible to instructors.
The AI detection tool identifies two types of content:
- AI-generated content — text created directly by large language models
- AI-paraphrased content — originally AI-generated text that was modified using paraphrasing tools
AI Score Color Codes and Symbols
| Indicator | Score Range | What It Means |
|---|---|---|
| Blue badge with % | 20–100% | Successfully processed; percentage reflects suspected AI content |
| Asterisk (*%) | 1–19% | AI may be present but confidence is low; numeric value hidden to prevent misinterpretation |
| Gray "--" | No score | Report couldn't generate — old submission, unsupported format, or scanned PDF |
| Red error (!) | No score | System error or processing failure — try resubmitting or contact support |
Note: As of July 2024, Turnitin no longer shows a numeric percentage for AI scores below 20%. Instead, these appear as an asterisk (*%) to prevent instructors from overreacting to low-confidence results.
In the detailed AI report, color-coded highlights show which parts of the paper were flagged:
- Cyan highlights — sections likely written directly by an AI model
- Purple highlights — sections likely generated by AI and then paraphrased through another tool
What's a "Good" Turnitin Score?
A "good" Turnitin score isn't defined by a fixed number — it depends on context. Here's a general framework:
Under 15–20%: Generally safe for most assignments. Common phrases and properly cited quotations typically account for this range.
15–30%: Often acceptable if the matches come from properly cited quotes and references. Worth reviewing the details to understand what's being flagged.
35%+: Warrants a closer look. Check whether the overlaps come from excessive quoting, poor paraphrasing, or missing citations. A high score doesn't automatically mean plagiarism, but it does need attention.
Something to watch for: A very low score (1–5%) might actually suggest the student didn't engage enough with external sources. In research-heavy assignments, some matching text from cited sources is expected and healthy.
Frequently Asked Questions
Why is my similarity score so high?
High scores often result from excessive quoting, weak paraphrasing, missing citations, or matching common academic phrases. Always review the report to see exactly what's being flagged before worrying.
How can I lower my similarity score?
Write in your own words, paraphrase effectively (don't just swap synonyms), cite all sources properly, reduce excessive quoting, and make sure your bibliography is formatted correctly so it can be filtered out.
Is 25% a bad Turnitin score?
Not necessarily. 25% is acceptable if the matches are from properly cited quotations and references. What matters is the nature of the matches, not the number itself. Check the detailed report to understand what's being flagged.
Is 40% similarity on Turnitin bad?
40% indicates a large portion of your paper matches other sources, which typically warrants review. If the matches are mostly from uncited material or poor paraphrasing, it could be problematic. If they're from properly cited quotes, it may be fine after filtering.
What does a 50% Turnitin score mean?
Half of your document matches other sources. This is a significant amount and usually indicates overreliance on external material, excessive quoting, or potential plagiarism. You should review the report carefully and consider revising.
Worried About Your Turnitin Score?
AuraWrite AI helps you polish AI-assisted drafts into natural, authentic writing that passes Turnitin's AI detection. 500 free words, no credit card required.