How to Read a Plagiarism Report

plagiarism-checker-online.net Editorial Team  |  Updated October 3, 2026

You have a number, say 17%, and a page full of highlighted text. Is that bad? Which highlights deserve your attention, and which are just your reference list doing what reference lists do? This guide explains what a report contains, sorts matches into five types with a clear action for each, and gives you a ten-minute routine you can apply to any report, from our scan or from your university's software.

What the Similarity Score Is and Isn't

The score is, in plain terms, the share of your text that matches other sources. Turnitin defines its similarity score as the percentage of text in a submission that matches other sources, and states that it doesn't check for plagiarism as such: it compares your text with its database and highlights similar passages so that a person can judge them. The library at Aberystwyth University puts it the same way, saying the software can only state how much text has been matched, not whether the work was plagiarized.

That has three consequences for how you read the number:

• It depends on the database. A tool can only match what it can see. Text from a paywalled article, a print book or a site that blocks automated reading may not appear. PlagAware, which powers our scans, states that it uses publicly available sources and has no access to paywalled articles, and that it can't guarantee detection of every public source.
• It depends on the settings. Whether quotes, the bibliography and very short matches count changes the figure. More on that below.
• It says nothing about intent or citation quality. Turnitin's own guide gives an example of one student who copies a chunk from a website and lands at 20%, and another who quotes and references several sources correctly and lands at 22%. The higher score belongs to the honest paper.

If your question is "what limit applies?", see acceptable plagiarism percentage. This article is about the next step, once you hold a report and have to decide what to do.

Anatomy of a Report

Layouts differ between tools, but nearly every report has the same four parts:

1. Overall score. One figure for the whole document. Useful for orientation, not for decisions.
2. Highlighted text. Your document with matched passages marked. This is where you do the actual reading.
3. Source list. The documents your text matched, usually with a link and the number or share of matched words per source.
4. Match detail. For each match, the original passage next to yours, so you can compare wording.

Our reports are generated by PlagAware: matched passages are highlighted and linked to the original sources, so you can open each source and compare the two texts. The how it works page shows the process from upload to report. Turnitin's student view adds a Match Groups panel that sorts matches by whether they have quotation marks and citations, which is the same logic used in the table below.

Direct matches vs. paraphrase matches

Some matches are word-for-word copies, others are the same sentence with a few words swapped or reordered. A good tool finds both, and both can be a problem when there is no citation. Don't rely on highlight colors to tell you how serious a match is: colors mean different things in different tools, and some use one color for everything. Read the passage and its source side by side, and judge it by what you see.

Five Types of Matches and What to Do

Almost every highlighted passage fits one of these five types. The examples are invented for illustration.

Types 4 and 5 are the ones that matter. If you aren't sure where the line between a paraphrase and a problem sits, read paraphrasing vs. plagiarism.

Step by Step: Triage in 10 Minutes

1. Filter out the noise, if your tool allows. Exclude the bibliography and, for a first pass, quoted text. This shows you what's left in your own prose.
2. Sort sources by size. Open the source with the most matched words first. One source covering 8% is more important than ten sources at 0.3% each.
3. Check each match against your notes. Did you take this wording from the source while reading? Do you have the citation? Most problems come from notes that mixed quotes and paraphrase.
4. Fix by type. Quote it, paraphrase it and cite it, add the missing citation, or cut it. Work from type 5 down to type 4.
5. Re-scan once. You're checking that your fixes didn't create new close matches, not hunting for a particular number.

A worked example

Below is a made-up report for a 2,400-word essay with an overall score of 18%. The three matches are invented to show the thinking.

Match B is the only one that needs work, even though match C is bigger. That's the whole point of reading instead of counting.

What to Exclude and How Settings Change the Score

The same document can produce different scores in different places. Turnitin's guides describe filters that let you hide quoted text and bibliographic material, and exclude individual matches or sources, so a report can be narrowed to what's relevant. An instructor's configuration decides what you see: Turnitin's own documentation notes that even correctly quoted and referenced text still appears as a match unless a filter hides it.

So a score from one tool isn't the score you'll get from another. A self-check is for finding passages that need work, not for predicting your university's number. Our guide on checking your paper before Turnitin goes deeper into that difference.

Thesis and Dissertation Notes

Long documents add a few wrinkles:

• Self-matches. If you scan chapters separately or submit drafts, later versions can match earlier ones. Reuse of your own earlier writing is its own question, covered in what is self-plagiarism.
• Appendices and tables. Survey instruments, interview guides and copied forms can match standard sources. Check whether your program counts them.
• Literature review. It naturally produces many small matches, mostly titles and standard terms. Look for runs of matching sentences rather than isolated phrases.
• Scale. A 20,000-word thesis can show hundreds of matches. Work from the largest sources down and don't try to clear everything.

If you're working on a thesis, our plagiarism checker for theses covers how scanning long documents works.

When to Rescan and What Changed

Rescan after you've made substantive changes, not after each small edit. Here's what a before-and-after might look like for the made-up essay above, with illustrative numbers:

The drop matters less than what remains. A remaining 10% made of quotes with citations and references is fine. The same 10% made of uncited passages is not.

Frequently Asked Questions

What similarity score is too high?

There is no universal limit. Turnitin says it doesn't set a percentage you should aim for, and institutions or instructors that use a threshold usually treat it as a trigger for review, not a verdict. A 25% report made of correct quotations and references can be fine, while a 6% report with one copied page is a problem. Ask your instructor or program for their rule in writing, and read the matches. Typical ranges are listed in our acceptable plagiarism percentage guide.

Why is my bibliography flagged?

Reference entries are copies of published records: author names, titles, journals and DOIs appear in many papers and databases, so they match. That's expected and not plagiarism. Some tools let you filter bibliographic material out of the score, but your instructor's settings decide what they see, so don't assume the filter is on.

Can a plagiarism report be wrong?

Yes, in both directions. It can flag common phrases, correctly cited quotations and your own earlier work, and it can miss sources it can't see, such as paywalled articles, print books or pages that block automated reading. Treat the report as a list of leads to check, not a ruling.

Does the report show AI-written text?

No. A plagiarism report shows overlap with existing sources. Text written by an AI tool can have a very low similarity score because it is newly generated, not copied. Detecting likely AI writing is a separate analysis, which we offer as the AI Scan, and its result is an estimate to review, not proof.

How long does it take to get a report?

Our scans usually deliver the report in about 15 minutes. Upload your file, choose the scan type, pay, and the report arrives by email. Pricing is $0.29 per page for the Plagiarism Scan, with a $0.90 minimum order, and one page is 1,800 characters including spaces. See the plagiarism checker page for details.

Sources

• Turnitin Guides: Understanding the similarity score
• Turnitin Guides: Student overview of the new Similarity Report experience (match groups, filters and exclusions)
• Turnitin Guides: Managing filters and exclusions in the Similarity Report
• Aberystwyth University LibGuides: How to interpret your Turnitin similarity report
• PlagAware: Frequently asked questions (public sources only, no access to paywalled articles; page in German)

This article is part of our Plagiarism Checking Guide.