Direct answer
There is no single official "good" number, because Turnitin does not set one. In practice, most institutions treat a similarity score of roughly 20% or lower as low, read alongside the source list rather than in isolation [1]. A "good" report is one where every highlighted match is properly quoted, cited, or a common phrase, not one that simply shows a small percentage [2].
What Similarity Percentage Is Considered Good or Acceptable on a Turnitin Report?
The honest starting point is that Turnitin deliberately avoids publishing a pass/fail threshold, because the similarity score measures matched text, not plagiarism [1]. The percentage tells you how much of your document overlaps with material in the comparison database; it says nothing about whether that overlap is legitimate [1]. A 30% score built entirely from correctly quoted sources and a reference list can be perfectly fine, while a 12% score containing one uncited copied paragraph can be a serious problem.
What most institutions do instead is apply informal bands. Turnitin's own guidance groups scores into colour ranges, and a score below roughly 20% is commonly read as low, with green, yellow, orange, and red bands signalling increasing levels of matched text [2]. Many departments treat 20% or under as acceptable for undergraduate work, while some postgraduate or journal contexts expect lower, and some technical subjects with heavy standard terminology tolerate higher [2].
The critical habit is to stop treating the number as a verdict and start treating it as a map. Turnitin allows instructors to exclude quotes, bibliographies, and small matches, so the same essay can legitimately display two different percentages depending on the settings used [2]. If your score looks high, check what is being counted before you panic or celebrate.
How Do You Read the Similarity Score and Matched Sources on a Turnitin Report?
The report is built around two linked elements: the overall percentage and a numbered source list, with matched text highlighted directly in your document [3]. Every highlighted passage is colour-coded to a source, so you can click a match to see which source it came from, how much of your text matched, and whether the system recognised it as a quotation [3]. Reading the sources is what actually tells you whether a match is a problem.
When you inspect a match, ask three questions. Is this text quoted and cited? Is it a common phrase, a title, or a standard method description? Or is it my own paraphrase that drifted too close to the original? A match that survives all three checks as "quoted and cited" is not something you need to fix, even if it pushes your total upward [3].
It also helps to know that instructors can filter the view. Excluding quotes, bibliography, and matches under a word-count threshold changes the displayed score, which is why two people looking at the same submission can report different numbers [3]. If your instructor quotes a figure that differs from yours, the exclusion settings are usually the reason, not a discrepancy in the submission itself.
How Can You Check Your Own Draft's Similarity and AI Score Before You Submit?
The most useful thing you can do with a similarity report is treat it as feedback rather than a final grade. Where students have access to their own report, the recommended approach is to review matches early, fix missing or incorrect citations, and reduce over-quoting before the final version goes in [4]. That turns the report into a revision tool instead of a source of last-minute anxiety.
Reviewing your own draft also surfaces the patterns that quietly raise scores: long block quotes with no analysis around them, paraphrases that keep the original sentence structure, and reused phrasing from your own earlier assignments [4]. Because Turnitin compares against a large repository, your own previously submitted work can appear as a match, which is why reusing your own paragraphs without attribution is still flagged [4].
A pre-submission check lets you see both signals at once. Similarity catches matched text; the AI writing indicator flags passages that read as machine-generated, and these are separate measures that can point in different directions [1]. Seeing both before you submit gives you time to revise the actual weak passages rather than guessing which number matters.
If reading your own report still leaves you guessing whether a score is safe, the fastest way to settle it is to run your draft through the same kind of check your instructor will use. turnitin0 produces a genuine Turnitin AI detection report and similarity report from a single upload, so you can see the real score, the flagged passages, and the matched sources before the deadline rather than after it.
※ Turnitin0.com - Actual Turnitin AI Report Cover, Score, Flag And Similarity Summary
FAQ
Is 15% similarity bad on Turnitin?
No. Turnitin does not define any score as bad or good, and a score around 15% is generally read as low [1]. What matters is whether the matched passages are quoted, cited, or common phrasing [2].
Can I have 0% similarity and still be flagged for plagiarism?
Yes. A 0% score only means no text matched the comparison database, which can happen if you paraphrased uncited ideas or translated a source [1]. Similarity and originality are not the same thing.
Does Turnitin's AI score appear on the same report as the similarity score?
They are separate indicators. The similarity score measures matched text, while the AI writing indicator flags machine-generated patterns, and a document can score low on one and high on the other [1].
Why does my instructor's percentage differ from mine?
Exclusion settings are the usual cause. Instructors can exclude quotes, bibliography, and small matches, which changes the displayed score without changing your submission [3].
Should I aim for a specific number before submitting?
Aim for a report where every match is explained by a citation, a quote, or common phrasing, rather than chasing a target percentage [4]. If you want certainty, checking your own draft first shows you the same matches your instructor will see [4].