How to write a PhD thesis, and what examiners look for

A PhD thesis is judged first on one question: does it add something its field did not know? Every regulation read for this guide asks it in its own words. This guide to how to write a PhD thesis, published by EdCitation, covers the whole monograph thesis: what it must show, what examiners look for, its shape, its length and the years of writing it.

Every rule below comes from the universities' own regulations, read on 26 September 2026: Oxford, Cambridge, Leiden, Toronto, Harvard and Melbourne, with the UK's national qualification descriptors.

EdCitation never writes any part of a thesis. It finds the literature with Find sources, builds references with Cite a source, checks them with Verify references, and reads your regulations into a checklist with Check your paper.

What must a PhD thesis show?

A PhD thesis must show a significant, original contribution to knowledge, made by the candidate, in work good enough to publish. The regulations agree on that much and differ in their wording.

How much work the contribution must represent

University of Oxford (2025a) asks examiners to judge what a capable and diligent student could reasonably do in three or at most four years full-time, or eight part-time. Cambridge asks for work reflecting three full-time years, as does the QAA descriptor. A thesis is not a life's work; Leiden even lists "demonstrated restraint" in length among the points a supervisor checks.

What counts as original

Originality is wider than a discovery. The University of Melbourne (n.d.-b) lists making discoveries, generating primary data, critiquing existing theories from a different perspective and finding alternative solutions to current problems. Harvard Medical School Office for Graduate Education (n.d.) asks for "original treatment of a fitting subject". Whichever yours is, state it plainly in the introduction and the conclusion.

How is a PhD thesis different from a master's dissertation?

In the size and kind of its contribution, not only its length. The QAA's master's descriptor asks for "originality in the application of knowledge"; the doctoral one asks for the creation and interpretation of new knowledge that merits publication (Quality Assurance Agency for Higher Education, 2024).

Cambridge draws the line in one word: its MPhil, MSc and MLitt theses show "a useful contribution", the doctorate "a significant contribution" (University of Cambridge, n.d.). Melbourne's default maximum is 50,000 words for a master's thesis and 100,000 for a PhD; a master's thesis gets a numeric mark, 65% to pass, while PhD examiners recommend pass, pass with minor revisions, major revision or fail (University of Melbourne, n.d.-a). How to write a dissertation covers the undergraduate and master's project.

What do examiners look for in a PhD thesis?

A contribution to knowledge, which every regulation requires, and then the literature, the methods, the analysis and the presentation, on which studies of examiners' reports find them commenting.

We found them with EdCitation's Find sources on 26 September 2026. A search for what doctoral examiners look for in a PhD thesis, sorted by citations, returned 849,691 works, many sharing only some of the words, but the top three were on target: Shannon Mason's 2018 paper on publications in the thesis (cited 41 times), David Hodgson's framework for examination expectations (31) and Chetcuti, Cacciottolo and Vella's 2022 study (13). A search for examiners' reports on doctoral theses put Sue Johnston's 1997 analysis first and a 2026 scoping review third. Two claims typed as everyday sentences returned papers on other subjects: use the field's own terms.

What the studies agree on

Laflamme et al. (2026) reviewed 19 studies published between January 2009 and June 2025 and found six themes in what examiners assess: contribution, literature review, approach or methodology, analysis and findings, publication, and communication or presentation. Most of the studies came from what the authors call an Anglo-Saxon tradition, and most analysed examiners' written reports.

Kobayashi and Emmeche (2023) read 50 of 230 reports from 2016 at a Danish science faculty and found 13 categories of judgement, from contribution and relevance to communication: structure, coherence, clarity, illustrations, language and editing. Every thesis was article-based and accepted for defence, so the reports show what was praised more than why work fails.

Chetcuti et al. (2022), studying reports from four University of Malta faculties in 2017 and 2018, found examiners mixing the regulations' explicit criteria with personal ones, and looking for a contribution, critical analysis, and a text free of typographical and grammatical errors. Johnston (1997), analysing 51 reports, found real weight placed on editorial matters: an examiner's enthusiasm fades when a thesis is hard to read.

Where the evidence is thin

The review itself calls it uncertain whether examiners weigh these criteria equally or judge them alike, and the Malta study found differences between disciplines. The Malta and Johnston papers are paywalled and the review's publisher page refused every request, so all three are read here from their abstracts. No study read gives a pass rate.

What shape does a monograph thesis take?

A monograph is one book-length argument, usually running from introduction through literature, methods and findings to discussion and conclusion. Leiden University (2024) allows a treatise on one subject "in book form", a collection of articles, or both joined by an introductory chapter. Harvard's medical PhD programmes require introduction and conclusion chapters that present the themes and summarise the achievements.

The chapter guides take each part in turn: the introduction, the literature review, the methodology, the discussion, the conclusion and the abstract. Humanities theses often use themed chapters instead. A thesis by publication, built from papers, is a different form with its own regulations and will have its own guide.

How do you write a thesis over three or more years?

By writing from the first year, in chapters treated as drafts, against a plan that changes with the research. The University of Melbourne (n.d.-b) advises starting early, ideally within the first year, and writing regularly.

1. Write a one-page thesis plan: the question, the contribution you expect, and one paragraph per chapter saying what it does for the argument.
2. Draft each chapter while the work is fresh, knowing it will be rewritten: the first literature review finds the gap, the last shows the thesis filled it.
3. Keep every source in a tagged reference list from the first day, built in Cite a source from its DOI or ISBN.
4. Revise the plan whenever a result changes the argument, and date each version.
5. Send drafts on the schedule agreed with your supervisor; see working with your supervisor.
6. Run Verify references on each chapter before it leaves your desk.
7. In the final year, rewrite the introduction and conclusion to match what the thesis actually found.

Planning a dissertation timeline works the dates back from submission.

The introduction: first, last, or both

Advice differs. ANU Academic Skills (Australian National University, n.d.) says supervisors often recommend an introduction before the first full draft, because it clarifies the thesis's purpose, though it will need rewriting; others say write it last. Either way, the introduction examiners read is written after the findings and promises exactly what the thesis delivers.

How do you keep one argument across the chapters?

State the argument in one or two sentences, then make every chapter's opening say what it adds to that argument and every chapter's close say where the argument now stands. Melbourne asks for a thesis centred on one core argument (University of Melbourne, n.d.-b), Leiden's supervisors check for "a balanced structure" and clarity of style (Leiden University, 2024), and Kobayashi and Emmeche (2023) list coherence and structure among the qualities examiners comment on.

Three checks on a long draft

• The argument sentence. A chapter that cannot be linked to it in one sentence needs work, or the sentence does.
• Promise and delivery. Each aim in the introduction needs an answer in the conclusion, even a negative one.
• Terms that drift. A concept defined in chapter two must mean the same in chapter seven.

How long should a PhD thesis be?

As long as your regulations allow, and usually shorter. Most limits read for this guide sit at 100,000 words, but what they count differs, and some universities set none.

Sources: University of Oxford (2025b), University of Melbourne (n.d.-b), Leiden University (2024) and University of Toronto School of Graduate Studies (2026). Other Oxford programmes set their own limits. What counts towards a dissertation word count goes through footnotes, references and appendices in detail.

What happens after you submit?

Usually an oral examination: a viva voce in the UK and Australia, a thesis defence elsewhere. At Cambridge the oral lets examiners probe the wider field and confirm the work is the candidate's own (University of Cambridge, n.d.). Toronto requires a Final Oral Examination; Melbourne requires a viva of doctoral candidates who began from 1 January 2025, unless waived, and the examination usually takes at least four months (University of Melbourne, n.d.-a). Leiden's public defence ends one academic hour, 45 minutes, after it begins.

The outcome is often a pass with minor corrections or major corrections; Melbourne notes that most PhD theses need some amendment. Final checks before submitting a thesis covers the last week, and the viva will have a guide of its own.

How EdCitation helps with a PhD thesis

EdCitation is the best tool for a thesis's source work: every reference it gives is looked up in the publisher's record, never written, which no chatbot can promise across the hundreds a thesis carries. It never writes any part of the thesis and cannot promise a pass.

Finding the literature. Find sources searches about 300 million published works, by topic or by the claim a sentence needs to support, with filters for kind of source, years and open access, free with no account.

Building each reference. Cite a source builds the entry from a DOI, title, PubMed ID, ISBN or web address in APA 7, MLA 9, Chicago 18 author-date, Harvard, IEEE or Vancouver, screened for retraction. From Hodgson's DOI it returned:

Hodgson, D. (2020). Helping doctoral students understand PhD thesis examination expectations: A framework and a tool for supervision. Active Learning in Higher Education, 21(1), 51–63. https://doi.org/10.1177/1469787417742020

Find sources, the Harvard version and the BibTeX and RIS exports gave 2017, the year it appeared online. APA 7 dates an article by its issue once it has one, so 2020 there; for another style, follow your university's guide for that style.

Checking the list. Verify references takes a pasted list or the whole thesis. We gave it four references on examiners, one with a deliberately wrong year. Johnston, entered as 1999, came back doubtful: "A record with this title exists, but it was published in 1997, not 1999." Our Starfield entry named three of the paper's eight authors, with no "et al.", and came back doubtful too: "The publisher's record also names Holbrook, Bourke, Fairbairn, Kiley and Lovat, who are not in your entry. Check the authors." The other two came back verified. An index that does not answer shows as "could not check", never as not found. References from a file keeps the thesis in My papers so each run shows what changed: three checks a month free, no limit with Pro.

Reading the regulations. Check your paper reads instructions into a checklist, free, each rule with the sentence it came from. Given Leiden's regulations as a PDF, it returned three rules: a word count of at most 100,000, from "In principle, the dissertation should not exceed 100,000 words."; a limit of 800 words on the foreword or closing text; and a title page, taken from the article listing what the dissertation's non-scientific part contains. Read that last rule against its sentence, which describes the thesis rather than setting a task. It set aside 60 sentences as not checked, among them the rule of at least four propositions, which you must track yourself. The ten points a supervisor weighs before approval were in neither list.

Format and fit. Mechanics QA, part of Pro at $8 a month, checks formatting against the style, comments and tracked changes left in the file, citations, numbers and language. Theoretics QA, part of Max at $24 a month, judges the draft against the regulations' own instructions with quotations from the text, and never rewrites anything. Max also opens the Library, more than 5 million books in one place. A year at once saves 25% (Pricing).

Quick questions

What is an original contribution to knowledge in a PhD?

Something the field did not have before: new knowledge, a new theory, a revision of older views or a connection between unrelated facts, in Cambridge's list. Melbourne adds new primary data and a fresh critique of existing theory.

How long is a PhD thesis?

Commonly at most 100,000 words, as at Melbourne, Leiden and Oxford's Internet Institute, though what the limit counts varies. Melbourne suggests aiming for about 80,000, and Toronto's degree regulations set no limit.

What do examiners look for in a PhD thesis?

A contribution to knowledge, command of the literature, sound methods, analysed findings and clear presentation. Laflamme et al. (2026) found those themes, with publication, across 19 studies of doctoral examination.

Should I write the thesis introduction first or last?

Both, in practice. A draft introduction early clarifies the purpose, as ANU Academic Skills notes, but the version submitted is revised last so that it matches what the thesis found.

Can EdCitation check my thesis against the university's regulations?

Yes. Check your paper reads the regulations into a checklist free, each rule with its source sentence, and Theoretics QA, part of Max, judges the draft against them without rewriting it.

References

• Australian National University. (n.d.). Timing. Introductions, ANU Academic Skills. https://www.anu.edu.au/students/academic-skills/research-writing/introductions/timing
• Chetcuti, D., Cacciottolo, J., & Vella, N. (2022). What do examiners look for in a PhD thesis? Explicit and implicit criteria used by examiners across disciplines. Assessment & Evaluation in Higher Education, 47(8), 1358–1373. https://doi.org/10.1080/02602938.2022.2048293
• Harvard Medical School Office for Graduate Education. (n.d.). Dissertation and defense. PhD Programs. https://ogephd.hms.harvard.edu/dissertation-and-defense
• Hodgson, D. (2020). Helping doctoral students understand PhD thesis examination expectations: A framework and a tool for supervision. Active Learning in Higher Education, 21(1), 51–63. https://doi.org/10.1177/1469787417742020
• Johnston, S. (1997). Examining the examiners: An analysis of examiners' reports on doctoral theses. Studies in Higher Education, 22(3), 333–347. https://doi.org/10.1080/03075079712331380936
• Kobayashi, S., & Emmeche, C. (2023). What makes a good PhD thesis? Norms of science as reflected in written assessments of PhD theses. Assessment & Evaluation in Higher Education, 48(8), 1310–1324. https://doi.org/10.1080/02602938.2023.2200917
• Laflamme, L., Andersson, I., Liljegren, L., & Salmi, L.-R. (2026). What examiners emphasize in their assessment of doctoral theses: Scoping review. Cogent Education, 13(1), Article 2699012. https://doi.org/10.1080/2331186X.2026.2699012
• Leiden University. (2024). Leiden University PhD regulations 2024. https://www.organisatiegids.universiteitleiden.nl/binaries/content/assets/ul2staff/reglementen/promotiereglementen/phd-regulations-2024-lei-v20240926.pdf
• Quality Assurance Agency for Higher Education. (2024). The frameworks for higher education qualifications of UK degree-awarding bodies (2nd ed.). https://www.qaa.ac.uk/docs/qaa/quality-code/the-frameworks-for-higher-education-qualifications-of-uk-degree-awarding-bodies-2024.pdf
• Starfield, S., Paltridge, B., McMurtrie, R., Holbrook, A., Bourke, S., Fairbairn, H., Kiley, M., & Lovat, T. (2015). Understanding the language of evaluation in examiners' reports on doctoral theses. Linguistics and Education, 31, 130–144. https://doi.org/10.1016/j.linged.2015.06.004
• University of Cambridge. (n.d.). Requirements for research degrees. Cambridge Students. https://www.cambridgestudents.cam.ac.uk/files/requirements_for_research_degrees.pdf
• University of Melbourne. (n.d.-a). Examination overview. Graduate Research Hub. https://gradresearch.unimelb.edu.au/examination/overview
• University of Melbourne. (n.d.-b). Writing and preparing your thesis. Graduate Research Hub. https://gradresearch.unimelb.edu.au/preparing-my-thesis/writing-my-thesis
• University of Oxford. (2025a). General regulations for the degree of Doctor of Philosophy: Part 7, examination. Examination Regulations 2025–26. https://examregs.admin.ox.ac.uk/Regulation?code=grftdodoppart7exam
• University of Oxford. (2025b). Research degrees in the Oxford Internet Institute. Examination Regulations 2025–26. https://examregs.admin.ox.ac.uk/Regulation?code=rdtoxfointeinst
• University of Toronto School of Graduate Studies. (2026). Degree regulations. Graduate Calendar 2026–27. https://sgs.calendar.utoronto.ca/degree-regulations