Vancouver referencing is a numbered citation system widely used in medicine, healthcare, biomedical sciences, clinical research, and related scientific disciplines.
Instead of identifying sources in the text using an author’s surname and publication year, Vancouver normally uses numbers to connect statements with entries in a numbered reference list.
For example:
Antibiotic resistance is an increasing global health concern.¹
or, depending on the particular Vancouver implementation:
Antibiotic resistance is an increasing global health concern [1].
The corresponding reference appears in the reference list:
- Author AA. Article title. Journal Title. 2025;12(3):100-108.
The exact presentation of Vancouver citations can vary between institutions, journals, and publishers. The specific requirements of the university, journal, or organisation should therefore always take priority.
This guide explains the major principles of Vancouver referencing, including numbered citations, journal articles, books, websites, clinical research, systematic reviews, guidelines, reports, datasets, and reference-list auditing.
Vancouver is a numerical referencing system associated particularly with:
The system is designed to keep citations relatively compact.
Instead of writing:
(Smith, 2025)
a Vancouver paper may use:
(1)
or:
[1]
or a superscript number:
¹
The precise form depends on the required Vancouver variant.
The system is associated with the International Committee of Medical Journal Editors and its historical meeting in Vancouver.
Over time, numerical referencing conventions became widely adopted by medical and scientific publications.
Today, however, the term “Vancouver style” can refer to a family of closely related practices rather than one perfectly identical format used everywhere.
This distinction matters.
The central principle is simple:
A source is assigned a number when it is first cited.
That number identifies the source throughout the document.
For example:
Clinical trials have evaluated several treatment approaches (1).
Later:
Similar findings have been reported in other populations (1).
The source remains reference number 1.
Suppose a paper first cites:
They become:
- Study A
- Study B
- Study C
The reference list follows this numerical order.
It is not normally organised alphabetically by author surname.
If a source is first cited as reference 4, it remains reference 4 throughout the paper.
For example:
First citation:
Previous research identified a significant association (4).
Later:
The same study also reported an important limitation (4).
Do not assign the source a new number merely because it appears again.
Depending on the implementation, citations may appear as:
(1)
or:
[1]
or:
¹
The assignment or publication instructions determine which form to use.
Do not mix citation formats in the same document unless the applicable style explicitly permits it.
Some Vancouver implementations use superscript numbers.
For example:
Smoking is associated with numerous health risks.¹
A later citation might be:
Similar findings have been reported elsewhere.²
If the same source is cited again:
The association has also been demonstrated in longitudinal studies.¹
The number remains unchanged.
Some institutions use bracketed numerical citations:
Smoking is associated with numerous health risks [1].
The reference list then contains:
- Author AA. Article title. Journal. Year;volume(issue):pages.
The exact punctuation depends on the required version.
Another implementation may use:
(1)
For example:
Several studies have examined this relationship (1-4).
Again, the institution or publication’s instructions determine the exact format.
A common problem is combining:
[1]
with:
²
and:
(3)
in the same document.
Choose the required presentation and apply it consistently.
The reference list should correspond directly to the citation numbers.
For example:
- Smith AB. Article title. Journal. 2025;10(2):100-108.
- Jones CD. Book title. London: Example Publisher; 2024.
- Brown EF. Report title. Example Organization; 2023.
Reference numbers correspond to the order of first citation.
Unlike many author-date systems, Vancouver generally uses citation order.
For example:
- Brown
- Smith
- Jones
is perfectly possible if Brown was cited first.
Do not alphabetise the list unless the specific requirements instruct you to do so.
When conducting research, it is useful to maintain a temporary source table.
| Source | First citation | Reference |
|---|---|---|
| Study A | 1 | 1 |
| Study B | 2 | 2 |
| Study C | 3 | 3 |
This becomes particularly valuable in long medical or scientific papers containing many sources.
Multiple sources can support one statement.
For example:
Several studies have reported similar findings.¹⁻⁴
or:
Several studies have reported similar findings [1-4].
Use the exact punctuation required by the selected Vancouver implementation.
If sources are not consecutive, cite the numbers separately.
For example:
Similar associations have been reported previously.¹,³,⁷
The exact punctuation depends on the style version.
Place the citation immediately after the claim, sentence, or information it supports.
For example:
The treatment reduced symptom severity in the study population.¹
This is clearer than placing a citation several paragraphs away from the relevant claim.
The reader should be able to determine which information a citation supports.
Weak:
The treatment is effective, inexpensive, widely available, and suitable for every patient.¹
The source may support only one of these claims.
Better:
The clinical trial reported improved symptom control in the study population.¹ The treatment’s cost and availability may depend on the healthcare setting.
This avoids overextending the source.
Do not assume that one citation at the end of a long paragraph automatically supports every statement in that paragraph.
If different claims come from different sources, cite them separately.
For example:
Study A reported an improvement in treatment response.¹ Study B found a different effect among older patients.²
This makes the evidence relationship clear.
Direct quotations should be used carefully in scientific and medical writing.
When exact wording is necessary, use the quotation format required by the institution or publication and identify the relevant source location when appropriate.
For example:
The researchers described the intervention as “highly effective” (1, p. 45).
Do not rely heavily on quotations when your purpose is to analyse scientific evidence.
Paraphrasing involves expressing information from a source using your own wording and structure.
For example:
Source idea:
The study found that early intervention reduced the duration of hospitalisation.
Paraphrase:
Patients who received earlier intervention experienced shorter hospital stays (1).
The citation remains necessary.
Changing a few words while keeping the original sentence structure is not an effective academic paraphrase.
A responsible paraphrase should:
Journal articles are among the most frequently cited sources in Vancouver-style medical writing.
A basic structure is:
Author AA, Author BB. Article title. Journal Title. Year;volume(issue):pages.
For example:
Smith AB, Jones CD. Example article title. Example Medical Journal. 2025;12(3):100-108.
Check the journal’s specific instructions before submission.
Large medical studies can have many authors.
The treatment of large author lists can vary.
Some implementations list a specified number of authors followed by:
et al.
Other publications may require a different treatment.
Always follow the current requirements of the journal or institution.
“Et al.” means “and others.”
It is used to shorten long author lists where the applicable style permits it.
For example:
Smith AB, Jones CD, Brown EF, et al.
Do not assume that every Vancouver implementation uses the same author cutoff.
Journal titles may be abbreviated in Vancouver-style references.
Medical publications frequently use standard journal abbreviations.
Use an accepted abbreviation rather than inventing one.
If a journal provides an official abbreviation, use it where required.
A journal article reference may include:
2025;12(3):100-108.
This generally represents:
The exact presentation may vary by source and publication requirements.
Online journal articles should still be treated as journal articles rather than generic webpages.
Where available, capture:
The DOI can provide a stable way of identifying the article.
A DOI can be particularly useful for identifying an electronic journal article.
Example structure:
Author AA. Article title. Journal. 2025;12(3):100-108. doi:10.xxxx/xxxxx.
Use the exact DOI supplied by the publisher.
Never invent a DOI.
Some journals do not use conventional page ranges.
Instead, an article may have an article number or e-location identifier.
In such cases, use the information supplied by the journal.
Do not invent page numbers.
Some articles are published electronically before being assigned to a final journal issue.
If citing an online-first article, use the publication information actually available and update the reference if the article later receives final volume, issue, and page information when necessary.
A basic book structure is:
Author AA. Book title. Edition. City: Publisher; Year.
For example:
Smith AB. Introduction to Clinical Research. 3rd ed. London: Example Press; 2025.
Use the edition information when the book is not the first edition.
For a book with multiple authors, list the authors according to the applicable Vancouver requirements.
Example:
Smith AB, Jones CD, Brown EF. Research Methods in Health Sciences. 2nd ed. London: Example Press; 2025.
For a large author group, consult the specific style requirements.
If the individuals are editors rather than authors, identify them as editors.
Example structure:
Smith AB, Jones CD, editors. Book Title. London: Example Press; 2025.
Do not describe editors as authors if the publication identifies them as editors.
A chapter can be referenced using information such as:
Example structure:
Smith AB. Chapter title. In: Jones CD, editor. Book Title. London: Example Press; 2025. p. 100-115.
A government agency, research institute, hospital, or professional organisation may be the author.
For example:
World Health Organization. Report Title. Geneva: World Health Organization; 2025.
Do not invent an individual author when the organisation is responsible for the publication.
Government documents are often important sources in healthcare and public health research.
They may include:
Identify the responsible government body accurately.
Clinical guidelines can be highly important sources.
Before citing a guideline, record:
Do not confuse a clinical guideline with a primary research article.
Systematic reviews are often important evidence sources.
A systematic review should be cited as the publication it actually is.
For example:
Smith AB, Jones CD. Systematic review title. Journal. 2025;12(3):100-110.
Do not automatically cite every study included in the review when you are specifically discussing the review’s conclusions.
A meta-analysis should also be cited as the relevant research publication.
If you use a numerical effect estimate from a meta-analysis, the meta-analysis should be identified as the source.
For example:
The pooled analysis found a significant reduction in risk (5).
Randomised controlled trials are primary research sources.
When discussing trial results, cite the actual trial report where possible.
For example:
The randomised trial reported a lower incidence of the primary outcome in the intervention group (8).
Avoid citing a general review when the specific trial result is the evidence being discussed.
Cohort studies can provide evidence about:
Cite the original cohort study when discussing its findings.
Case-control studies may be cited when discussing:
Be careful not to describe an association found in a case-control study as proof of causation.
The citation should support the precise claim.
Cross-sectional studies provide information about a population at a particular point or period.
If citing prevalence estimates from such a study, ensure the citation actually corresponds to the population and period being discussed.
Vancouver referencing can also be used in qualitative research.
Relevant sources may include:
The citation system remains numerical even though the research methodology is qualitative.
Case reports can be valuable for unusual clinical presentations.
However, their evidence has limitations.
When citing a case report, distinguish:
Do not use a single case report to make a universal clinical claim.
Clinical trial registries can provide information about:
When the registry record is relevant, cite the appropriate record according to the applicable Vancouver requirements.
Research protocols can be useful when discussing study design or planned methodology.
A protocol is not automatically equivalent to a completed research study.
Make the distinction clear in your writing.
Medical databases can be useful for locating sources.
However, a database such as PubMed is primarily a discovery platform.
When you find an article through a database, cite the article rather than treating the database search page as the source.
PubMed can help researchers identify:
However, database metadata should still be checked against the original publisher record where possible.
Some medical sources have identifiers such as:
These identifiers can help verify a source.
Do not automatically include every identifier unless the applicable reference style requires or permits it.
A basic online-source structure may include:
Organisation. Title of webpage [Internet]. Location: Publisher; date [cited date]. Available from: URL
The exact treatment varies between Vancouver implementations.
Follow the instructions of the institution or journal.
Online information can change.
For this reason, Vancouver references may include a cited or accessed date.
For example:
[cited 2026 Sep 15]
Use the required date format consistently.
If a webpage has no reliable publication date, do not invent one.
Use the appropriate treatment for undated material according to the selected Vancouver guide.
An access date may still be important.
If an organisation is responsible for a webpage, it may serve as the corporate author.
If neither an individual nor an organisation can reasonably be identified, follow the applicable Vancouver guidance rather than inventing an author.
News articles can sometimes be relevant to public health or policy research.
However, distinguish journalism from peer-reviewed research.
A news article should not automatically be used as evidence for a scientific claim when the underlying study is available.
Government websites can be valuable sources for:
Prefer the official government source when it directly provides the information.
Professional bodies can publish:
These can be authoritative sources when directly relevant.
Check the date and version.
Reports can be cited when they provide relevant evidence.
A report reference may contain:
Use the information actually provided by the report.
A thesis or dissertation can be cited as an academic source.
Example structure:
Author AA. Thesis title [dissertation]. City: University; Year.
The exact degree designation and format depend on the Vancouver implementation.
Conference proceedings may be useful sources, especially for emerging scientific research.
A conference reference can include:
Follow the specific requirements of the reference guide being used.
Abstract-only sources should be used cautiously.
An abstract may not provide enough information to evaluate:
If the full article is available, prefer the full article.
Preprints can provide access to emerging research before formal peer review.
If citing a preprint, identify it clearly according to the applicable Vancouver requirements.
Check whether a peer-reviewed version has subsequently been published.
Datasets should be cited when they are directly used in research.
A dataset citation may include:
Use the dataset’s recommended citation whenever one is provided.
Software should be cited when it materially contributes to the research.
Examples include:
Record the version used.
If the results depend on a specific software package, documenting the software version can improve reproducibility.
For example:
Statistical analyses were performed using software version 5.2.
The relevant software should be cited if the applicable guidelines require it.
A medical device may need to be identified by:
Manufacturer documentation can be cited when specifications materially affect the research.
When discussing a drug, distinguish between:
For important therapeutic claims, prefer authoritative medical evidence.
If discussing a trial of a particular medication, cite the original trial publication where possible.
A later review can be used for synthesis, but it should not obscure the underlying evidence.
When citing a clinical recommendation, determine whether the recommendation comes from:
The source type matters because each represents a different level and type of evidence.
A strong research paper should consider the quality and relevance of evidence.
Depending on the research question, useful evidence can include:
Citation style does not determine evidence quality.
A citation supports a statement; it does not automatically establish causality.
For example:
Study X found an association between exposure A and outcome B (4).
This does not necessarily justify:
Exposure A causes outcome B.
The strength of the conclusion must match the study design.
When reporting:
cite the study that produced the estimate.
Do not alter the numerical result during paraphrasing.
If you report a statistical finding from another study, preserve:
Cite the original source.
A clinical statement should be supported by an appropriate source.
For example:
The intervention reduced hospitalisation in the reported trial (7).
This is preferable to an uncited statement about clinical effectiveness.
Be especially careful with statements involving:
Use language that accurately reflects the evidence.
Systematic reviews often contain numerous citations.
When writing a review:
A literature review should not become a list of numbered citations.
Instead:
Citations support each stage of the discussion.
In a research proposal, Vancouver citations can establish:
Use recent and authoritative evidence where appropriate.
The research question should emerge from the evidence and identified gap.
For example:
Previous studies have investigated treatment A in adults (1-4), but limited evidence is available regarding its use in the target population (5,6).
This can provide a logical basis for a research question.
Objectives normally do not require citations simply because they are objectives.
However, the background leading to those objectives should be supported by appropriate evidence.
Cite established methods where they are adopted from published research.
For example:
The sampling procedure followed the approach described by Smith et al. (4).
If the method is your own design, explain it without pretending that an external source created it.
If you use a validated questionnaire, scale, diagnostic instrument, or assessment tool, cite the original development or validation study where appropriate.
This establishes the basis for using the instrument.
For a validated scale, identify:
Do not assume that a scale is validated for every population simply because it has been used elsewhere.
Established statistical methods may be cited when a specific published methodology is central to the analysis.
For routine statistical procedures, citation requirements may depend on the institution.
Follow the methodology guidance for the discipline.
Research involving human participants may require discussion of:
If an ethical framework or guideline is cited, reference the authoritative source.
Relevant sources can include:
Use the actual document governing the research.
Public health research frequently uses:
Ensure that statistics are cited to the source that actually produced them.
When presenting:
cite the original statistical source.
Also verify the:
Do not present an old statistic as though it were current.
If the research depends on current prevalence or incidence, verify the publication date and data period.
Citation formatting cannot correct outdated evidence.
International health studies often use reports from organisations such as:
Use the authoritative source whenever possible.
Case studies may use several source types:
Citations should distinguish external evidence from the author’s case analysis.
Nursing assignments may use Vancouver because nursing programmes often use numerical medical referencing systems.
Common sources include:
Follow the university’s nursing-school requirements.
Pharmacy research may require references to:
Primary evidence should be used where appropriate.
Dental research and assignments may use Vancouver-style citations.
Potential sources include:
Check the dental school’s specific requirements.
Biomedical research often contains large numbers of scientific sources.
The numbered format helps maintain readability in technically dense papers.
However, the same fundamental rule applies:
Every citation should correspond to reliable evidence.
Laboratory papers may cite:
Cite the source when a method or specification comes from external work.
Genetics research may require references to:
Record version information for databases and software where relevant.
Biotechnology papers can involve:
Each source should be classified and cited appropriately.
Research involving artificial intelligence in healthcare may combine:
Do not cite a technical AI paper as evidence of clinical effectiveness unless it actually investigated clinical outcomes.
For medical AI research, distinguish between:
A model’s high accuracy on a dataset does not automatically establish clinical effectiveness.
Citations should support the specific claim being made.
Health data science papers may cite:
Documenting data and software sources supports reproducibility.
If a database itself is a research resource, it may require a citation.
Examples include:
Use the database’s recommended citation if available.
Preclinical studies may involve:
Cite the original research and relevant methodological sources.
Avoid generalising preclinical results directly to humans.
When discussing animal research, identify the actual study and accurately represent:
Do not generalise beyond the evidence.
Case reports can provide useful clinical observations but generally have limited generalisability.
When citing them, use them for the specific evidence they provide.
Editorials and commentaries can provide professional perspectives.
However, distinguish opinion or commentary from empirical research.
Do not cite an editorial as though it were a clinical trial.
Letters can contain:
If the letter is relevant, cite it according to the journal’s bibliographic information.
A retracted article should not normally be treated as ordinary evidence.
Before citing an important study, especially in medical research, check whether it has been:
This is a source-quality issue rather than merely a formatting issue.
If a published article has a correction, verify whether the correction affects the information you are citing.
Where necessary, cite the corrected record appropriately.
Medical evidence can change rapidly.
For topics such as:
verify whether more recent evidence has superseded older findings.
A source may have been authoritative when published but no longer represent current evidence.
When writing current clinical or scientific work:
Search engines are useful for finding sources.
They are not normally the sources themselves.
Instead:
Google Scholar can help locate:
However, automatically generated citations should be checked manually.
Metadata errors can occur.
Online citation generators can help with formatting.
They should be treated as assistance rather than authoritative sources.
Always verify:
Reference managers can help with:
But they do not guarantee correct references.
After generating a Vancouver bibliography, manually inspect a sample of references from:
If errors are common, inspect the entire bibliography.
Incorrect:
Brown
Jones
Smith
Correct principle:
- First source cited
- Second source cited
- Third source cited
Incorrect:
First citation = 3
Later citation = 14
Correct:
First citation = 3
Later citation = 3
If the text contains:
(1), (2), (4)
investigate why reference 3 is missing.
A missing number may indicate a numbering error.
Every numbered citation should correspond to the correct reference.
If citation [8] points to the wrong article, the entire evidence trail is compromised.
Do not invent abbreviations.
Use an accepted journal abbreviation where the selected Vancouver format requires one.
Do not rearrange authors alphabetically.
Preserve the author order supplied by the publication unless the applicable reference style specifically instructs otherwise.
If the source contains a relevant page range and the reference format requires it, include it.
Do not invent page numbers for electronic articles that use article identifiers.
PubMed is a database.
If you found an article through PubMed, cite the article itself.
Do not cite the PubMed search results page as though it were the journal publication.
If you are discussing the exact result of a clinical trial, cite the trial itself where practical.
Use the review to discuss the broader synthesis.
Clinical recommendations can change.
Before citing a guideline, verify:
A blog may provide useful general information, but it should not automatically replace:
AI-generated text is not automatically a reliable scientific source.
If AI was used during research or writing, follow the institution’s AI policy.
Do not cite an AI-generated claim as though it were a peer-reviewed study.
Instead, locate and cite the underlying authoritative evidence.
Never invent:
A fabricated reference is an academic integrity problem.
Correct referencing helps demonstrate that:
Citation is therefore part of responsible academic research.
Plagiarism can involve:
A citation should accompany appropriate paraphrasing or quotation.
Previously submitted work may also require careful handling.
Do not assume that your own previous academic work can always be reused without disclosure.
Follow the institution’s rules on:
AI tools may assist with:
However, the researcher remains responsible for verifying sources.
AI tools can generate:
Every AI-generated reference should therefore be independently verified.
A strong Vancouver bibliography should reflect the quality of the underlying research.
Aim for:
Do not pursue a high reference count simply for appearance.
A medical paper may contain many citations.
However, citation quantity is not the same as evidence quality.
Weak:
Many researchers have studied this problem (1-20).
Stronger:
Randomised trials have reported reductions in the primary outcome (1-4), while observational studies have produced more variable estimates (5-8).
The second formulation tells the reader what the sources contribute.
A strong literature review can use citations to compare:
This turns a bibliography into an evidence-based argument.
Weak:
Study A found X (1). Study B found Y (2). Study C found Z (3).
Stronger:
Study A reported improved outcomes in a younger population (1), whereas Study B observed weaker effects among older participants (2). This difference may reflect variation in patient characteristics and study design. A third study reported similar results after adjusting for baseline risk factors (3).
The citations support analysis rather than replacing it.
A good Vancouver literature review should answer:
Citation numbers make the evidence traceable.
For a long dissertation:
A major issue is citation numbering across chapters.
If the thesis uses one continuous reference list, citation numbers should normally remain consistent across the document.
Do not restart numbering in every chapter unless the institution specifically requires chapter-specific references.
Appendices may contain:
External material used in an appendix may still require citation.
Check the institution’s formatting rules.
Tables containing external evidence should identify their sources.
For example:
| Intervention | Outcome | Source |
|---|---|---|
| Intervention A | Improved | (3) |
| Intervention B | No significant change | (7) |
If an entire table is derived from one source, a table note may identify the source.
External figures should be appropriately attributed.
For example:
Figure 2. Disease progression model adapted from reference 5.
Do not assume that changing colours or labels removes the need for attribution.
If you substantially modify a published figure, table, or conceptual model, identify it as adapted where appropriate.
This is both an attribution and intellectual-property consideration.
Citation and copyright are related but not identical.
A citation acknowledges the source.
It does not necessarily grant permission to reproduce copyrighted material.
For substantial reproduction, check applicable copyright and permission requirements.
Open-access material can be cited like other scholarly sources.
However, open access does not automatically mean:
Evaluate the source itself.
Before citing a journal article, consider:
Do not assume that a journal is authoritative merely because it appears online.
For important research, check whether a source has been retracted.
This is particularly important in medical research because a retracted clinical study can materially affect conclusions.
For each important source, ask:
A useful reference database can contain:
| Field | Example |
|---|---|
| Author | Smith AB |
| Title | Example article |
| Journal | Example Medical Journal |
| Year | 2025 |
| Volume | 12 |
| Issue | 3 |
| Pages | 100-108 |
| DOI | 10.xxxx/xxxxx |
| PMID | 12345678 |
| Source type | Journal article |
| Key finding | Reduced risk |
| Relevant page | 104 |
This can make final citation checking much easier.
For complex research, map major claims to sources.
Example:
| Claim | Source | Citation |
|---|---|---|
| Disease prevalence | National report | 1 |
| Treatment effectiveness | Clinical trial | 2 |
| Safety evidence | Systematic review | 3 |
| Clinical recommendation | Guideline | 4 |
This reduces unsupported statements.
A final citation audit should check:
Are they sequential according to first appearance?
Do repeated citations retain the same numbers?
Does every number point to the correct source?
Are the sources appropriate?
Is the information accurate?
Are URLs and access dates correct?
Are current guidelines and evidence used where necessary?
Check every reference for:
Only include information that is actually available and required.
If you add a new source near the beginning of the paper, all later citation numbers may shift.
For example:
Original:
Study A = 1
Study B = 2
Study C = 3
After inserting a new source first:
New Study = 1
Study A = 2
Study B = 3
Study C = 4
Reference management software can automate this.
Manual numbering becomes risky when:
Use a reference manager or automated citation system where practical.
When using Microsoft Word, citation tools can help manage references.
However, imported citation metadata should still be checked.
Do not assume that a generated bibliography is automatically correct.
LaTeX can provide strong automated reference management for long scientific documents.
The bibliography database should still be checked for:
Automation manages formatting; it does not guarantee accurate data.
In group research projects, agree on:
This prevents different team members from using different reference formats.
International research teams may have members accustomed to:
Choose one required system for the final document and standardise the entire manuscript.
| Feature | Vancouver | APA | IEEE |
|---|---|---|---|
| Main citation system | Numerical | Author-date | Numerical |
| Reference order | Citation order | Alphabetical | Citation order |
| Common fields | Medicine/health | Social sciences | Engineering/technology |
| Repeated source | Same number | Author-date repeated | Same number |
| Citation example | (1) | (Smith, 2025) | [1] |
The exact formatting can vary by implementation.
Vancouver and IEEE are both numerical systems, but they should not be treated as identical.
They may differ in:
If an assignment specifically requires Vancouver, use Vancouver rather than substituting IEEE.
The major difference is the citation mechanism.
Vancouver:
(1)
APA:
(Smith, 2025)
Vancouver is particularly compact for documents containing many citations.
APA makes author and year information immediately visible.
Harvard uses an author-date system.
Vancouver uses numerical references.
For example:
Vancouver:
Previous research found an association (3).
Harvard:
Previous research found an association (Smith 2025).
Vancouver is appropriate when:
Do not select Vancouver merely because you prefer numbered citations.
Do not use Vancouver when the assignment explicitly requires:
The required academic style takes priority.
For student work:
For a journal article:
A medical research paper may contain:
Citations are particularly important in the Introduction and Discussion because these sections establish the relationship between the current research and existing evidence.
Use citations to establish:
Avoid filling every sentence with citations when the evidence is already adequately established.
Cite:
Explain modifications to existing methods.
Your own research results generally do not need citations.
However, if you compare your results with external findings, cite the relevant studies.
For example:
The observed prevalence was higher than that reported in previous studies (5,7).
The Discussion often requires substantial citation.
Use references to:
The conclusion should primarily synthesise the study’s findings.
Do not introduce large amounts of new evidence in the conclusion unless the document structure specifically permits it.
Citation requirements for abstracts vary.
Many journals discourage or prohibit references in abstracts.
Always follow the publication’s instructions.
Some journals impose limits on the number of references.
If a journal has a reference limit:
Do not remove critical evidence simply to meet an arbitrary target without considering the argument.
Recent evidence can be important in rapidly evolving fields.
However, “recent” does not automatically mean “better.”
A foundational study may still be essential.
Use a balanced source strategy.
A seminal study may be decades old but remain important.
For example:
The original diagnostic framework was introduced in an earlier study (1).
Cite the foundational source when the historical development matters.
For clinical recommendations and rapidly changing topics, check whether more recent evidence has changed earlier conclusions.
Use updated systematic reviews and guidelines where appropriate.
When studies disagree, do not hide the disagreement.
For example:
Several studies reported improved outcomes (1-3), whereas two later investigations found no statistically significant difference (4,5).
This provides a more accurate evidence picture.
When discussing:
cite methodological sources when appropriate.
A strong paper distinguishes evidence from interpretation.
If a study reports an association, consider whether confounding variables could explain it.
Do not present an observational association as proof of causality without appropriate evidence.
If you report a sample size from another study, cite the study.
Do not accidentally confuse:
When discussing limitations of previous research, cite the original study and accurately represent what the researchers reported.
Do not invent limitations simply because they seem plausible.
A convincing research gap should be supported by the literature.
Avoid statements such as:
No one has studied this topic.
unless you have sufficient evidence to justify such a broad claim.
Prefer carefully qualified language:
Limited research has examined this question in the target population.
When conducting a systematic or structured literature search, maintain records of:
Citation management is only one component of rigorous evidence synthesis.
During literature screening, distinguish:
Reference-management software can help manage these categories.
The same research may appear as:
Determine which version should be cited.
Avoid accidentally treating multiple versions as independent studies.
Sometimes research is published in more than one language or journal.
Check whether the publications represent:
Do not double-count the same evidence.
Good referencing should allow readers to understand:
This is especially important in health research.
Where relevant, document:
Good references support reproducibility.
If public data are used, cite the dataset or official data source.
Also state the relevant:
If your study analyses data originally collected by another organisation, cite the dataset or source.
Explain the relationship between:
If a questionnaire or scale is borrowed or adapted:
If using a translated publication, accurately identify the source and translation where required.
Do not silently treat a translated work as though it were originally published in the language you are using.
Non-English research can be important.
If the source is relevant and authoritative, cite it according to the applicable Vancouver rules.
Where a translated title is included, follow the required bibliographic convention.
Do not rely exclusively on international studies when a research question concerns a specific population.
Local evidence may be important for:
Use relevant local research where available.
For research concerning African populations, consider evidence from:
Do not assume that evidence from another population applies identically.
When applying findings from one population to another, discuss relevant differences.
These can include:
Citations should support the evidence, while the researcher should make the applicability judgment explicit.
A reliable Vancouver workflow is:
| Source | Typical reference information |
|---|---|
| Journal article | Author, article title, journal, year, volume, issue, pages |
| Online journal article | Author, title, journal, year, volume, pages/article number, DOI |
| Book | Author, title, edition, location, publisher, year |
| Book chapter | Chapter author, chapter title, editor, book, pages, publisher, year |
| Website | Author/organisation, webpage title, publication details, cited date, URL |
| Guideline | Organisation, title, version/year, publisher, URL where applicable |
| Government report | Organisation, report title, publisher, year |
| Thesis | Author, title, degree, institution, year |
| Conference paper | Author, title, conference, location/date, publication details |
| Dataset | Creator, dataset title, version, repository, year, identifier |
| Software | Creator/organisation, software name, version, year |
| Clinical trial | Trial authors, article title, journal, publication details |
| Systematic review | Review authors, title, journal, publication details |
| Standard | Organisation, standard title/number, year |
Before submitting a Vancouver-referenced paper:
No. Vancouver is primarily a numerical citation system.
Yes. Sources are normally assigned numbers according to their first appearance.
Yes. A source normally retains its original citation number throughout the document.
Generally, no. The reference list normally follows citation order.
Yes. Some implementations use superscript numbers.
Yes. Some institutions and publications use bracketed numerical citations.
No. Both use numerical systems, but their detailed reference conventions differ.
No. APA uses an author-date system, while Vancouver uses numerical citations.
Yes, when they are relevant and reliable sources.
Yes. Clinical guidelines are often important sources in medical and health research.
Yes. A dataset should be cited when it contributes directly to the research.
Yes. Software should be cited when it materially contributes to the research.
Normally, no. PubMed is primarily a database for locating biomedical literature. Cite the underlying publication.
When discussing a specific study’s findings, the original study is generally preferable. Reviews are valuable for synthesis and broader context.
AI can assist with formatting and organisation, but every reference must be independently verified.
Use this guide alongside other Lamtas academic writing resources:
Lamtas provides ethical academic support intended to help learners and researchers strengthen their own work.
Relevant services include:
Support can include:
The objective is to support learning and improve the quality of learner-developed work rather than misrepresenting someone else’s work as the student’s own.
Vancouver referencing is designed to make source attribution compact and traceable, particularly in medical, biomedical, clinical, and health-science writing.
The essential principles are:
The strongest Vancouver-referenced paper is not simply one with correctly numbered references. It is a paper in which reliable sources are selected carefully, evidence is represented accurately, claims are supported appropriately, and readers can trace the research behind the author’s conclusions.