How do I reference in Vancouver style? NLM rules with real examples

What is Vancouver referencing?

Vancouver is the numbered citation style most biomedical journals use outside the US, and many inside it. The name comes from a 1978 meeting of medical journal editors in Vancouver. That meeting produced the "Uniform Requirements for Manuscripts Submitted to Biomedical Journals", whose reference formats came from the US National Library of Medicine (NLM). The document is now the ICMJE Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals, last updated in January 2026.

Today the ICMJE doesn't write the reference rules itself. It refers authors to two NLM sources:

• NLM's Samples of Formatted References for Authors of Journal Articles: one page with more than 40 numbered examples covering articles, books, websites, software and other sources.
• Citing Medicine: The NLM Style Guide for Authors, Editors, and Publishers, 2nd edition: the full manual, free on the NCBI Bookshelf.

So "Vancouver style", "NLM citation style" and "ICMJE referencing style" all describe the same family of rules. Individual journals then add their own variations (brackets or superscript, number of authors, whether to include the month).

The core rules:

1. Number references consecutively, in the order they are first mentioned in the text.
2. Mark citations in the text, tables and legends with Arabic numerals. The ICMJE says to put them in parentheses.
3. Give a source the same number every time you cite it.
4. List the references at the end in numerical order, not alphabetically.
5. Abbreviate journal titles as in the NLM Catalog.

How do you write a Vancouver reference for a journal article?

The NLM pattern is:

Author AA, Author BB. Title of article. Abbrev Journal. Year Mon Day;Volume(Issue):pages. doi: 10.xxxx

Every example below is a real article, checked against its PubMed record. The month is included when PubMed has it.

One author

Ioannidis JP. Why most published research findings are false. PLoS Med. 2005 Aug;2(8):e124. doi: 10.1371/journal.pmed.0020124.

Two authors

Walters WH, Wilder EI. Fabrication and errors in the bibliographic citations generated by ChatGPT. Sci Rep. 2023 Sep 7;13(1):14045. doi: 10.1038/s41598-023-41032-5.

Four authors: list them all

Bhattacharyya M, Miller VM, Bhattacharyya D, Miller LE. High rates of fabricated and inaccurate references in ChatGPT-generated medical content. Cureus. 2023 May;15(5):e39238. doi: 10.7759/cureus.39238.

More than six authors: first six, then et al. (journal convention)

Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021 Mar 29;372:n71. doi: 10.1136/bmj.n71.

Older article without a DOI

Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986 Feb 8;1(8476):307-10.

Formatting points:

• No italics or bold anywhere. Journal names are abbreviated but plain.
• Page ranges are shortened: 307-10, not 307-310.
• "et al." has a period and follows a comma after the last listed author.
• Article titles go in sentence case, as in the NLM samples. PubMed's Cite button keeps the journal's original capitalization (for example "High Rates of Fabricated..."), so fix it by hand when your target journal wants sentence case.
• If the journal numbers its pages continuously through a volume, NLM lets you leave out the month and issue: Lancet. 1986;1:307-10.
• Adding the PMID or PMCID at the end is optional in NLM style. Include them only if the journal asks for them.

Do you list all authors in Vancouver, or six then et al.?

Both rules appear, and that is why people get confused:

In practice most journals that use Vancouver ask for the first six authors and then et al. Some ask for three, and a few want every author. Set your reference manager to the journal's rule rather than editing each entry by hand.

How do you reference books, chapters, websites and DOIs in Vancouver?

A few rules from Citing Medicine that people miss:

• Book titles go in sentence case and are not italicized.
• The place of publication comes before the publisher. Add the state or country in parentheses when the city is not well known: Chichester (UK), Bethesda (MD).
• If you cannot find the place, write [place unknown]; for a missing publisher, [publisher unknown].
• Online sources need a [cited ...] date, the date you looked at them.
• Most journals accept the short print form plus "doi: 10.xxxx" for online articles, which is what PubMed generates. The longer [Internet] form is the full Citing Medicine version.

How do you do Vancouver in-text citations?

Put the number in the text where you use the source.

Parentheses are the ICMJE default. Many journals use square brackets [5] or superscript⁵ instead. The ICMJE doesn't say where the number goes relative to punctuation, so copy your target journal's practice and use it the same way throughout.

What's the difference between Harvard and Vancouver referencing?

Is APA the same as Vancouver? No. APA is an author-date style like Harvard. For a full side-by-side of AMA, Vancouver and APA, see our AMA citation style guide.

How do I insert Vancouver citations in Word, EndNote or Zotero?

Use a reference manager and let it number and format the list. Then check its output against the journal's rules.

1. EndNote: in Word, open the EndNote tab and choose the Vancouver output style from the Style menu, or use "Select Another Style" to find it or a journal-specific style. Insert citations with Cite While You Write. If the journal wants superscript or a different author limit, edit a copy of the style instead of each reference.
2. Zotero: install the Word or LibreOffice plug-in. Under Document Preferences, pick "NLM/Vancouver: Citing Medicine 2nd edition (citation-sequence)". If it is not on the list, add it from the Zotero Style Repository, which also has styles for many individual journals.
3. Mendeley and other managers: most include a Vancouver or NLM style. Look for a journal-specific style first.
4. Word's own References tab: go to References, then Style, and see whether your version of Word offers Vancouver. The built-in style list is fixed and fairly short, so most medical authors use one of the plug-ins above.
5. Quick single references: the Cite button on a PubMed record produces an NLM-format citation you can paste. Fix the capitalization and trim the author list to the journal's rule.

A "Vancouver citation generator" can only format what you give it. If the metadata is wrong, you get a nicely formatted wrong reference. AI chatbots are worse: published studies have found fabricated and inaccurate references in ChatGPT-generated medical content. Check every reference against PubMed or Crossref before you submit. CliniPaper does this automatically and flags anything it cannot match.

Worked example: cleaning a reference list (synthetic draft)

Here is a synthetic draft entry of the kind that appears when references are pasted from several sources:

3. Page, Matthew J. et al., "The PRISMA 2020 Statement: An Updated Guideline for Reporting Systematic Reviews," British Medical Journal 372 (2021): n71.

What needs fixing for Vancouver:

1. Authors: surname plus initials, with no periods or commas inside a name. List up to six (if the journal uses the six-author rule), then ", et al."
2. Title: sentence case, no quotation marks.
3. Journal: the NLM abbreviation "BMJ", no italics.
4. Date and volume: Year Mon Day;Volume:pages.
5. Add the DOI.

Corrected:

3. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021 Mar 29;372:n71. doi: 10.1136/bmj.n71.

If you are writing a systematic review, the reference list has to agree with the studies in your PRISMA flow diagram. Reviewers do check.

Common mistakes reviewers flag

If a reviewer raises reference problems, list each correction in your response to reviewer comments so the editor can see that the whole list was checked.