The plagiarism limit for a PhD thesis in India is 10% similarity, because the UGC 2018 regulations treat up to 10% as Level 0 with no penalty, and almost every PhD scholar hears it as “keep it under 10%”. That number is real, but it is only a small part of the rule. The actual regulation says more about which text counts than about the percentage, and much of the concern around similarity reports comes from reading a raw score before the filters are applied.
This guide sets out what the UGC (Promotion of Academic Integrity and Prevention of Plagiarism in Higher Educational Institutions) Regulations, 2018 actually say, taken from the Gazette notification itself. It covers the four levels, the penalties for theses and for publications, the exclusions, how universities layer their own rules on top, and how to read a DrillBit, Turnitin or Ouriginal report without fooling yourself in either direction.
What the UGC 2018 plagiarism regulations say
The UGC notified the Regulations on 23 July 2018, and they were published in the Gazette of India on 31 July 2018. They apply to students, faculty, researchers and staff of all higher educational institutions in the country, and cover “scripts”: research papers, theses, dissertations, book chapters and books submitted for Master’s and research-level degrees or for publication.
Several duties fall on the institution, not on you. Under Regulation 6, each university must adopt a technology-based checking mechanism, frame a plagiarism policy approved by its statutory bodies and put it on its website. It must also require two documents with every thesis: an undertaking from the scholar that the work is original and has been checked with a tool approved by the institution, and a certificate from the supervisor that the work is plagiarism free. After the degree is awarded, the thesis goes to INFLIBNET’s Shodhganga repository within a month — which matters later, because open-access theses such as those on Shodhganga can be part of what checkers compare against.
The four levels and the penalties for a thesis
Regulation 8 quantifies plagiarism into four levels by similarity percentage. Regulation 12.1 attaches a penalty to each level for theses and dissertations. The percentages are the regulation’s own wording; the penalties are paraphrased closely.
| Level | Similarity | Penalty for a thesis or dissertation (Reg. 12.1) |
|---|---|---|
| Level 0 | Up to 10% | Minor similarities. No penalty. |
| Level 1 | Above 10% to 40% | The student is asked to submit a revised script within a stipulated period not exceeding six months. |
| Level 2 | Above 40% to 60% | The student is debarred from submitting a revised script for one year. |
| Level 3 | Above 60% | The student’s registration for that programme is cancelled. |
Three details in Regulation 12 are easy to miss:
- Repeat offences move up a level. A scholar found plagiarising again is punished at one level higher than the previous finding.
- A degree already awarded is not safe. If plagiarism is proved after the award, the degree can be put in abeyance for a period recommended by the Institutional Academic Integrity Panel and approved by the head of the institution.
- Penalties follow a process, not a number. Penalties are imposed only after misconduct is “established without doubt”, all avenues of appeal are exhausted and the person has had a fair opportunity to defend themselves. A Departmental Academic Integrity Panel investigates and reports within 45 days; an Institutional panel reviews and decides.
The same four levels apply to faculty and researchers for publications, with different penalties:
| Level | Penalty for academic and research publications (Reg. 12.2) |
|---|---|
| Level 0 (up to 10%) | No penalty. |
| Level 1 (above 10% to 40%) | Asked to withdraw the manuscript. |
| Level 2 (above 40% to 60%) | Withdraw the manuscript; denied one annual increment; may not supervise new Master’s, M.Phil. or Ph.D. scholars for two years. |
| Level 3 (above 60%) | Withdraw the manuscript; denied two successive annual increments; may not supervise new scholars for three years. |
What is excluded from the count
This is the part of the regulation that most changes the number you see. Regulation 7 lists what the similarity check shall exclude:
| Excluded by Regulation 7 | What it means in practice |
|---|---|
| Quoted work reproduced with all necessary permission and/or attribution | A properly quoted and cited passage should not count. The checker cannot see permission, so quotation marks and a citation are what a reviewer looks for. |
| References and bibliography | Your reference list will match thousands of other papers. It should be excluded, either by the checker’s filter or by removing it from the file checked. |
| Table of contents, preface and acknowledgements | Front matter follows fixed wording. Some universities ask you to submit only the chapters for checking for this reason. |
| Generic terms, laws, standard symbols and standard equations | Names of statutes, standard formulae (Bernoulli’s equation, the chi-square statistic) and discipline vocabulary are not plagiarism. |
| Common knowledge or coincidental terms, up to 14 consecutive words | Stated in the note to Regulation 7. Short runs of ordinary phrasing are not counted. This is where the “small match” filter comes from. |
The note on “core work”, which is often overlooked
The note under Regulation 7 says a scholar’s work “shall be based on original ideas, which shall include abstract, summary, hypothesis, observations, results, conclusions and recommendations only and shall not have any similarities.” Read literally, the 10% is an overall figure, and the sections that carry your own contribution are expected to have essentially no copied text at all. A university may check these sections separately, so ask. In practice, similarity in the literature review and methods is normal; similarity in your results or conclusions chapter draws an examiner’s attention even if the overall score is low.
How universities apply the levels
Because each institution writes its own policy under Regulation 6, the same thesis can pass at one university and be returned at another. The differences are usually in these places:
- Which tool, and who runs it. Most universities accept only a report generated by the library or research office on the institution’s licensed tool. A report you ran yourself on a free website is not a substitute.
- What is submitted for checking. Some ask for the full thesis with the bibliography removed; others ask for chapters only; some also check the synopsis at the pre-submission stage.
- Filter settings. Whether quotes and bibliography are excluded automatically, the small-match threshold (the 14-word note is a natural reference point), and whether the scholar’s own earlier submissions to the repository are excluded.
- Stricter internal thresholds. An institution or department may set lower acceptable figures than 10%, or set a per-chapter limit. If your department has one, it applies to you regardless of the UGC minimum.
- Published papers from the thesis. Policies differ on whether your own published papers can be excluded as sources. This is covered below.
If your university’s format guide covers plagiarism checking, it is usually in the submission section. See our Anna University thesis format guide for an example of where these rules sit alongside layout requirements.
What individual universities’ documents say
These come from the university documents behind our thesis-format guides. Rules change, so check the current version with your research section.
| University | What its own documents say | Our guide |
|---|---|---|
| Manonmaniam Sundaranar | Up to 10% on DrillBit (Revised Ph.D. Regulations 2023, 11.4) | MSU thesis format |
| Alagappa | 10% on iThenticate, with references and your own papers excluded (Ph.D. Regulations, August 2026, 12.5) | Alagappa thesis format |
| Periyar | “Acceptable limit 20%” on the plagiarism certificate form (Annexure 14); the regulations themselves give no figure | Periyar thesis format |
| Annamalai | No percentage in the 2026-27 regulations outside the ICAR stream of the Faculty of Agriculture | Annamalai thesis format |
| Savitribai Phule Pune (SPPU) | Similarity of 10% or less, with a certificate (Amended Ordinance 186, 10.15) | SPPU thesis format |
| Mumbai | Up to 10% carries no penalty (Ph.D. Regulations 2026, R.PhD.31) | Mumbai thesis format |
| Kerala | No percentage or software named in the Ph.D. Regulations 2022 (issued 07.12.2023) | Kerala thesis format |
| Calicut | 10% on the university’s plagiarism check certificate; UGC 2018 regulations adopted by University Order | Calicut thesis format |
| Delhi | No percentage; the library lists DrillBit for checking theses and its forms refer to the UGC 2018 norms (Delhi University Library System instructions, posted May 2025) | Delhi thesis format |
| Osmania | 10% similarity and a separate 20% limit on AI-generated content (Ph.D. Rules 2026-27, 28.1) | Osmania thesis format |
DrillBit, Turnitin and Ouriginal: which checker you will meet
Under the Ministry of Education’s ShodhShuddhi programme, run by the INFLIBNET Centre, universities and institutions have had access to plagiarism detection software since 1 September 2019. INFLIBNET provided Ouriginal (formerly Urkund) until that contract ended on 30 September 2023. From 1 October 2023 it has provided DrillBit-Extreme, with full service from 1 November 2023. So if your university uses the national licence, your thesis is most likely being checked on DrillBit; our DrillBit guide explains how that check works and how to read its report.
Many institutions also hold their own Turnitin licence (Ouriginal is now part of Turnitin), and some journals run iThenticate, Turnitin’s publisher product. Different tools search different databases, so the same chapter can give noticeably different percentages on two checkers. Neither number is “wrong”; the one that counts is the tool and settings your university uses for submission.
How to read a similarity report
A similarity percentage is the share of your text that matches something in the checker’s database. It is not a plagiarism verdict. A careful reading goes in this order:
- Check the filters first. Look at whether quotes, the bibliography and small matches are excluded, and what the small-match threshold is. An unfiltered report can easily be several times higher than the filtered one for the same thesis.
- Look at the source list, not the headline. Sort by the largest single sources. One source contributing 6% is a very different problem from sixty sources contributing 0.1% each. The first is usually a section closely paraphrased from one paper; the second is usually common phrasing.
- Open each large match. Is it a quotation you forgot to mark? A standard definition or a statute? A methods description that follows a standard protocol? Your own published paper? Or genuinely borrowed prose?
- Look for your own earlier drafts. If a synopsis or an earlier version of the thesis was submitted to the same checker’s repository, it may match your current draft almost entirely. That is a settings problem for the research office, not plagiarism, but you need to raise it before the report goes to your supervisor.
- Check where the similarity sits. Matches concentrated in the literature review are expected. Matches in results, discussion or conclusions need attention whatever the total.
Your own published papers
Most PhD theses in India include material from the scholar’s own journal papers, and many universities expect them to. The checker will flag those papers as sources. The UGC Regulations do not specifically exclude them, so the handling is set by your university: some allow the scholar’s own publications to be excluded on declaration, others expect them to count. Whatever the rule, cite the papers in the relevant chapters and state in the thesis which chapters are based on which publications. Remember too that the journal usually holds copyright in the published version; reuse in a thesis is normally permitted, but check the publisher’s policy.
How to reduce similarity through better scholarship
When a report is genuinely high, the fix is almost always better scholarship rather than rewording. What works:
- Quote and cite where the exact wording matters (a definition, a legal provision), and keep quotations short.
- Synthesise, don’t summarise paper by paper. A literature review that walks through sources one at a time tends to echo each source’s phrasing. Grouping studies by theme and comparing them in your own argument lowers similarity as a side effect of being better writing.
- Describe standard methods briefly and cite the protocol instead of reproducing it.
- Cite your own earlier work as you would anyone else’s.
This is the work our similarity reduction support does: a report reviewed match by match, with chapters restructured by someone in your field so that the argument is yours. You remain the author throughout, and the ethics policy sets out how we work.
Checker workarounds count as misconduct
Scholars are still offered ways to “beat” the software: swapping Latin letters for look-alike Cyrillic ones, white-coloured characters between words, text pasted as images, automated synonym spinners, or machine translation into another language and back.
These fail in three ways. Checkers and research offices look for them: character substitution and hidden text show up as anomalies, and a thesis full of images of text is obvious to anyone who opens the file. Spun text reads unnaturally, and examiners notice sentences that no researcher in the field would write. And the attempt itself is evidence of intent: a report with a low score and a manipulated file is far more serious than a genuine high score, which usually only means rework. Under Note 2 of Regulation 12.1, a degree can be put in abeyance if plagiarism is proved even after the award.
AI-written text and AI detectors
The 2018 Regulations were written before generative AI tools were in common use and do not mention them. Many universities have since issued their own circulars, and both Turnitin and DrillBit now offer AI-writing indicators. Check your institution’s current position before you rely on any tool.
Two cautions pull in opposite directions. First, AI-generated text is not your work, and presenting it as your own raises the same integrity question as copying. And a paraphrasing tool that lowers a similarity score may raise an AI score. Second, AI detectors give a probability, not proof. They are known to produce false positives, particularly on formulaic academic prose and on writing by non-native speakers of English, and Turnitin itself advises that its AI score should not be the sole basis for action against a student. If your own writing is flagged, keep your drafts, notes and version history; they are the best evidence of how the text was produced.
Sources
FAQ
Questions scholars ask
Is 10% similarity the pass mark for a PhD thesis?
Up to 10% is Level 0 under the UGC Regulations: minor similarity with no penalty. It is not a pass mark in the sense that anything below it is automatically fine: matches in your results or conclusions can still be questioned, and your university may set a lower internal limit or check chapters separately.
Does the bibliography count towards the similarity percentage?
It should not. Regulation 7 excludes references, bibliography, table of contents, preface and acknowledgements. Whether that is done by the checker’s filter or by submitting the file without those pages depends on your university’s procedure, so ask your library or research office which they use.
My own published paper is showing as the top source. What do I do?
This is common, because most theses are built on the scholar’s own papers. The UGC Regulations do not specifically address it, so follow your university’s policy: some allow exclusion on declaration. In any case, cite the paper in the chapter and state which chapters draw on which publications.
Why do DrillBit and Turnitin give different percentages for the same chapter?
They compare against different databases and use different matching and filter defaults. Neither is more correct. The number that matters is the one produced by the tool and settings your university uses for thesis submission.
What happens if my thesis is above 10% at submission?
Under Regulation 12.1, similarity above 10% to 40% leads to a revised script within a period not exceeding six months; above 40% to 60%, a one-year bar on resubmission; above 60%, cancellation of registration. In practice, most universities check before formal submission, so a high score usually means you revise before the thesis goes further.
Can someone guarantee my thesis will come out under 10%?
Similarity scores depend on the checker and its settings, so a specific number can’t be fixed in advance, and a promised figure usually relies on mechanical rewording. The durable way to lower similarity is to rewrite with your own synthesis and citations, with a filtered report run on the same settings your university uses.
