Turnitin Score for a Dissertation or Thesis: What Is Acceptable?
There is no universal acceptable Turnitin score for a dissertation or thesis. Many universities and supervisors use a limit somewhere between 10% and 20% with the bibliography excluded, and some set their own written threshold. What examiners care about more is where the matching text comes from. Here is how to judge your thesis similarity and avoid the problems that only show up in long documents.
Typical dissertation similarity limits
| Similarity (bibliography excluded) | How supervisors usually read it |
|---|---|
| Under 10% | Comfortable for most theses |
| 10 to 15% | Normal, especially with a long literature review |
| 15 to 20% | Often acceptable, but the sources will be checked |
| Above 20% | Usually needs revision or an explanation |
These are rules of thumb, not Turnitin rules. Check your programme handbook or graduate school policy first: if it names a figure, that figure is the one that counts.
What examiners look at beyond the number
- Single large sources. One source above about 3 to 5% of a long thesis stands out.
- The literature review chapter. It normally has the highest similarity. Close paraphrasing chapter after chapter is the main risk.
- Methods sections. Standard descriptions of methods and instruments match other theses. Cite them and keep your own wording.
- Your own earlier work. Chapters submitted before as coursework, a proposal or a published article will match. Tell your supervisor in advance and cite your published papers.
Problems specific to long documents
File limits
Turnitin's similarity check accepts files under 100 MB and 800 pages. The AI writing detector accepts up to 30,000 words. Many theses are longer, so check them chapter by chapter, or split the file into parts below 30,000 words each.
Your proposal or drafts in the repository
If your proposal or earlier chapters were submitted to Turnitin with repository storage on, your final thesis matches them and the score jumps. Ask whether earlier submissions were stored, and ask for them to be excluded from the final report.
Tables, appendices and questionnaires
Survey instruments, consent forms and standard scales match other theses that used them. Many institutions accept this, but say so in your submission notes.
The AI score for a thesis
More universities now run the AI writing report on theses as well as similarity. A thesis with low similarity can still get a high AI score if sections were drafted with AI tools. If your institution has a policy on AI use, follow it and keep notes of how you used any tool. The AI detector now reads English, Spanish, Japanese, Arabic, Dutch, German and Turkish. See every language Turnitin AI detects.
A 24-page paper with both reports
Both screenshots below come from the same 24-page, 6,364-word paper. The similarity report found 1% overall similarity. The AI writing report flagged 51% of the qualifying text as likely AI-generated. Tap either one to see it full size.
Screenshots shown for illustration. Your own reports arrive as full PDFs in your dashboard.
Check your thesis before you submit
- Split your thesis into chapters or parts of up to 30,000 words if it is longer.
- Create an account and buy one slot per part. Buy 10 slots, get 1 free: every 10 slots come with 1 free.
- Upload each part and download both reports, usually within minutes.
Your document is not added to the student repository, so your final submission will not match your own check. See also high similarity from references.
Frequently asked questions
What Turnitin score is acceptable for a dissertation?
There is no universal figure. Many universities and supervisors use a limit between 10% and 20% with the bibliography excluded. Your handbook's figure is the one that counts.
Can Turnitin check a whole thesis at once?
The similarity check accepts files under 100 MB and 800 pages. The AI detector accepts up to 30,000 words, so long theses are checked in parts.
Will my proposal match my final thesis?
Yes, if your proposal was stored in the repository. Ask for it to be excluded from the final report.