Referencing is the cheapest part of a thesis to get right and one of the easiest places to lose marks. An examiner who finds three citations that do not match the sentences they are attached to will start checking the rest, and a reference list that cannot be trusted makes a reader wonder about the table on page 40 that nobody can verify.
What Vancouver is
Vancouver[4] is a numbered citation style used by most medical journals and by most Indian health universities for the postgraduate thesis. A number in the text points to one entry in a list at the end, and that list sits in the order the numbers first appear rather than alphabetically. The name is historical: a small group of general medical journal editors met in Vancouver in 1978 and agreed a common format for the manuscripts they received, and that group grew into the International Committee of Medical Journal Editors, which publishes its own account of that history. The convention is carried forward today in the Committee's recommendations on the conduct, reporting, editing and publication of scholarly work, which send authors to the detailed reference formats maintained by the United States National Library of Medicine.
Whether your number goes in superscript, in round brackets or in square brackets, and whether it sits before or after the full stop, is set by your university's template. There is no single correct answer, and marks are lost for inconsistency rather than for choosing one form over another. Find the template before you write the first citation.
Numbering in order of first appearance
The first source you cite is reference 1. The next new source is 2. When you cite source 1 again on page 30, it is still 1. A reference that appears first inside a table or a figure legend takes its number from the point in the text where that table or figure is first mentioned, which is the rule people most often get wrong when they move a table during revision. Cite consecutive ranges as a range rather than a list. The moment you renumber anything by hand, the list has begun to drift.
The parts of a journal reference
A journal article is built from six elements in a fixed order: authors, article title, journal title abbreviated in the National Library of Medicine's form, year, the volume with the issue number in round brackets, and the page range. The article title takes sentence capitalisation, with no quotation marks and no italics. Page ranges are usually contracted, so 134 to 139 becomes 134-9. The pattern is worth memorising as a shape rather than copying from an example:
- Journal article: Author AB, Author CD, Author EF. Title of the article in sentence case. Abbrev J Name. 2024;41(3):210-6.
- Book: Author AB. Title of the book. 4th ed. New Delhi: Publisher; 2023.
- Chapter in an edited book: Author AB. Title of the chapter. In: Editor CD, Editor EF, editors. Title of the book. 3rd ed. Mumbai: Publisher; 2022. p. 88-104.
- Web source: Name of the issuing organisation. Title of the page [Internet]. Place: Publisher; 2024 [cited 2026 Sep 20]. Available from: the full URL.
- Thesis or dissertation: Author AB. Title of the thesis [dissertation]. Place: Name of the university; 2021.
The bracketed Internet tag and the cited date are what make a web reference checkable, and they are the two things almost always missing from a first draft. Regulatory documents, national guidelines and registry entries are usually cited in this form, so if your thesis leans on a guideline, this is the pattern you will use most.
Six or more authors
The sample references published for the ICMJE recommendations list the first six authors and then et al. when a paper has more than six. This is not universal: the National Library of Medicine itself now records all authors, and individual journals differ, some asking for every name and some for three followed by et al. For a thesis, follow your university's template where it states a rule, and where it is silent use the first six and et al. throughout. What is actually assessed is that every entry in the list follows the same rule.
The reference you have not read
Do not cite a paper you have only seen quoted inside another paper. If the original is genuinely unobtainable, say in the text that you are relying on a secondary report and cite what you actually read. Examiners ask which source a particular claim came from, and the question is usually aimed at the claim that sounds most confident. Reading at least the abstract, methods and limitations of everything you cite is slower on the way in and very much faster in the viva.
Using a reference manager without letting it corrupt the list
Zotero, Mendeley and EndNote all produce a clean numbered list, and all three will produce a broken one if they are fed bad records or edited by hand. The failure is almost never the software; it is the library behind it.
- Import records from PubMed or from the publisher's own page, never by typing them, and never from a pop-up citation box without checking what it produced.
- Check every record on the day you import it: author initials, abbreviated journal title, year, volume, issue, pages, and whether the title has lost its capitalisation, its Greek letters or its superscripts.
- Keep one library for the whole thesis. Two libraries is how the same paper arrives twice with two different numbers.
- Never edit a formatted bibliography by hand in your word processor. The field regenerates, and your correction disappears at the worst possible moment.
- Before you print, convert the citations to plain text in a copy of the file, and keep the linked version as your working document so later edits still renumber correctly.
- Read the converted copy as a PDF before it goes to the binder. Superscript numbers and page ranges are where the formatting usually breaks.
The errors that actually get caught
- A number pointing at the wrong entry because a paragraph was moved late in the revision.
- Entries sitting in the list that are never cited anywhere in the text.
- The same paper twice, from two slightly different imported records.
- Journal titles written out in full in one half of the list and abbreviated in the other.
- A web reference with no access date, or a URL that no longer resolves.
- An in-press citation for a paper that was published two years ago.
- Accented characters and symbols rendered as question marks after a conversion.
When to do it
Set aside an hour with the reference list open beside the thesis before each draft goes to your guide, and check the first citation of every reference against the sentence it supports. It is the highest-yield hour in the whole submission, because these are the only errors in a thesis that can be found and fixed with complete certainty. Everything else in the document is a matter of judgement. This part is simply either right or wrong.