Your paper comes back with a similarity score of 23%, and the first thing you do is google whether that's bad. Here's the uncomfortable truth behind every answer you'll find: there is no universal number that separates "fine" from "in trouble." A 30% can be completely innocent, and an 8% can get you referred to a misconduct panel. What decides your fate isn't the percentage — it's what's underneath it.
That said, "it depends" is a useless answer on its own. So let's go through what the score actually measures, the rough ranges instructors are used to seeing, and how to tell whether your number is a problem.
What the similarity score actually measures
The similarity score is the percentage of your text that matches other sources in the databases Turnitin checks — published work, web pages, and papers submitted by other students. That's all it is: matching text. It is not a plagiarism verdict. Turnitin itself is explicit that the report doesn't determine plagiarism — it just shows matches, and a human decides what they mean.
This is why the number alone can't be "good" or "bad." A properly quoted paragraph with a citation is matching text. Your bibliography is matching text. The lab-report template your instructor handed out is matching text. None of that is misconduct.
The color bands (and what people assume they mean)
Turnitin groups scores into color bands: blue for no matching text, green for 1–24%, yellow for 25–49%, orange for 50–74%, and red for 75–100%. Students often read this as a verdict scale — green means safe, red means expelled. Instructors don't read it that way. The bands are just a sorting aid; plenty of markers open the green papers too.
In practice, most ordinary essays with normal quoting and referencing land somewhere under roughly 20%, so scores in that region usually don't raise eyebrows by themselves. Many universities publish informal guidance in that range — but it varies by institution, by department, and by assignment type. A literature review will naturally score higher than a personal reflection. If your institution states a threshold, that number beats anything a blog tells you, including this one.
When a low score is still a problem
Here's the case that surprises people: an essay scores 8%, and the student still ends up in a misconduct meeting. Why? Because the entire 8% was one continuous block — a paragraph lifted from a journal article with no quotation marks and no citation. One uncited match of a few sentences is a stronger signal of plagiarism than thirty scattered two-word matches ever could be.
Markers read the match list, not just the total. One big match from a single source is what they're trained to look for.
When a high score is fine
The reverse also happens constantly. Common reasons for a scary number that melts away under inspection:
- Quotes and bibliography. If the instructor didn't exclude them from the report, a heavily referenced paper can gain 10–15 points from its reference list alone.
- Templates and boilerplate. Assignment cover sheets, standard methods sections, legal or technical phrasing that can only be written one way.
- Common phrases. "The results of this study suggest that" matches thousands of papers. Scattered tiny matches are noise.
- Your own draft. If an earlier version of your paper was stored in Turnitin's repository, the final version can match it — sometimes at a catastrophic-looking percentage. This is the self-match trap, and it's why any pre-check must run in no-repository mode (we wrote about this in how to check your score before submitting).
How markers actually read the report
Knowing the reading order helps you audit your own report the same way. A typical pass looks like this:
- Glance at the overall score to calibrate — not to judge.
- Open the match list and look at the largest single match. Is it one source contributing 12%, or forty sources contributing fractions of a percent?
- Check whether big matches are quoted and cited. Matched text inside quotation marks with a citation is scholarship; the same text without them is a problem.
- Look at paraphrases that track a source too closely — same sentence structure with swapped synonyms is the classic "patchwriting" flag.
Lowering your score the legitimate way
If your report shows real issues, the fixes are unglamorous but reliable: put direct quotes in quotation marks with citations, rewrite paraphrases from your own understanding rather than editing the source sentence, cite every idea that isn't yours, and don't paste in the assignment brief or template text. What doesn't work: synonym-swapping tools and "similarity reducers," which tend to produce mangled prose that reads worse than the problem it hides — and often trips the AI writing detector instead.
So, is your score bad?
Run the audit above on your own report: if every significant match is quoted, cited, or boilerplate, a 25% is defensible. If there's one uncited block from a single source, a 9% isn't. The percentage is a headline; the match list is the story.
And if you're reading this before submission day — that's the right time. Seeing the actual report while you can still fix the matches is the whole game; a real Turnitin check shows you every matched source, item by item, about 10 minutes after you upload.