Journal manuscript references are formatted last and checked least. This guide, published by EdCitation, takes the reference list family by family: the ICMJE and Vancouver journals, APA 7 journals, the numbered styles of Nature, Springer and IEEE, and the rest. Every rule was read at its own source on 23 September 2026; where a source could not be read, the guide says so.
The family is the one the journal's instructions for authors name. Read its references section, then a recent article, because the two occasionally disagree. The evidence on how often published references are wrong is in the leader of this cluster, how to check your references before journal submission, and is not repeated here. Three of the checks below, the lookup of every entry, the retraction screen and the two-way match of citations to the list, are what EdCitation's free Verify references runs on an uploaded manuscript; the rest are the journal's own rules, and they are set out here family by family.
What do Vancouver and the ICMJE ask for?
A list numbered in the order of first mention, verified against PubMed or the original papers, with no retracted article cited as evidence. The International Committee of Medical Journal Editors (2026) publishes the recommendations most biomedical journals follow, updated in January 2026.
What the ICMJE says may and may not be cited
- Numbering. Consecutive, in order of first mention. A reference cited only in a table or figure legend is numbered from where the text first identifies that table or figure.
- Abstracts. Cite published articles rather than abstracts wherever possible.
- Retracted articles. Authors must check that no reference cites a retracted article, except to refer to the retraction; for MEDLINE-indexed articles, PubMed is the authoritative source.
- Personal communications. Avoided unless essential and unavailable from a public source, and then with written permission and confirmation of accuracy.
- Submitted but unaccepted work. Cited in the text as "unpublished observations", with written permission, and not listed.
- Preprints. The citation must make clear that the reference is a preprint.
- Verification. Against a bibliographic source such as PubMed, or the original documents.
The NLM's form, and its abbreviations
The National Library of Medicine (2026) keeps the sample references the ICMJE points to, based on its manual Citing Medicine and last updated on 2 June 2026. The journal article form lists the first six authors and then "et al.", and the sample with a DOI ends "338:a2752. doi: 10.1136/bmj.a2752", which is also the model for an article number in place of pages. Citing Medicine itself could not be read: the NCBI site answered with a verification page.
Journal titles are abbreviated as MEDLINE abbreviates them, and the ICMJE names the NLM Catalog as the place to look them up; the National Library of Medicine (n.d.) makes it searchable by title, abbreviation or ISSN. Since 1 March 2007 each abbreviation has been built from the ISSN Centre's abbreviated key title, with each word capitalised and punctuation and diacritics removed, so "Journal of neural transmission. General section" becomes "J Neural Transm Gen Sect" (National Library of Medicine, 2024). Cite a source in Vancouver builds the entry from a DOI or a PubMed ID; check its journal title against the NLM Catalog before submission.
What does APA 7 ask for in a journal article reference?
The DOI as a link whenever the article has one, the issue number always, an article number where the journal uses one, and every author up to 20 in the list, as the American Psychological Association (2025a) sets out on its journal article page, last updated in August 2025.
- The DOI. Included whenever the article has one, written as a https://doi.org/ link in every example; the APA page on this site adds no full stop after it. With no DOI, the reference ends after the page range, or with a URL that resolves for readers.
- Article numbers. Where the journal gives one instead of a page range, the word "Article" and the number take its place: "PLoS ONE, 13(3), Article e0193972".
- Missing parts. A volume, issue or page number the journal does not use is omitted. An accepted paper not yet published takes "in press" in place of the year, and "submitted for publication" never goes in the date element (American Psychological Association, 2025b).
- Authors. Up to and including 20 in the list; with 21 or more, the first 19, an ellipsis, and the final author (American Psychological Association, 2025b). In the text, three or more authors become the first author plus "et al." in every citation, including the first (American Psychological Association, 2019); when to use et al. has the exceptions.
- Retracted articles and abstracts. A retracted article is cited with its original details, then the retraction's year, journal, volume, issue and pages in parentheses. An abstract is cited only where it is the only source available, marked "[Abstract from ...]".
What do the numbered styles at Nature, Springer and IEEE ask for?
Each numbers by first citation, and each counts authors differently: Nature stops at five, IEEE at six, and Springer's numbered journals do not say.
Nature Portfolio
Nature's formatting guide (Nature, n.d.) numbers references in the order they appear across the text, tables, boxes, figure legends, Methods and Extended Data. An Article has up to about 50 references in the main text, not counting the Methods. Article titles are upright, journal titles italic and abbreviated, the volume bold. All authors are listed unless there are more than five, and then the first author with "et al." Conference abstracts, patents, preprints on "recognized servers", papers in press and datasets with a DOI may be cited; a paper in preparation is mentioned in the text and not listed.
Springer's two families
The Journal of Materials Science (n.d.) numbers in square brackets, asks for DOIs as full links, and abbreviates journal titles by the ISSN List of Title Word Abbreviations; its example entry, "S. Preuss, A. Demchuk Jr., M. Stuke, Appl. Phys. A 61, 33 (1995)", carries no article title at all. Scientometrics (n.d.) cites by name and year, alphabetises the list, and encourages APA 7's count of authors while keeping an "et al." the authors have used. Both list only works cited and already published or accepted, and keep personal communications and unpublished work in the text.
IEEE
The IEEE Reference Guide (IEEE, 2025), version 3.28.2025, puts each reference number in square brackets on the line of text; it does not state the order in one sentence, but its warnings about renumbering after an edit assume the order of first citation. Up to six authors are listed; with more than six, the first author and "et al." The DOI closes the entry as "doi: 10.1109/TBME.2011.2158315", and every reference ends with a full stop unless it ends with a URL. IEEE gives a form for what the others keep out: an arXiv preprint with its year and number, "unpublished", "submitted for publication", "to be published", and "private communication" with its month and year. Cite a source in IEEE follows the guide.
What about Elsevier, Chicago and Harvard journals?
Elsevier's rules are set per journal, and none could be read for this guide: ScienceDirect, which hosts each Guide for Authors, refused every request, including one from a browser, as did the author pages of The Lancet and Cell. Nothing about Elsevier's rules is reported here; read your target journal's guide.
A Chicago author-date journal wants "(Kucsko et al. 2013)", no comma before the year, and the 18th edition drops the place of publication on a book; Cite a source in Chicago follows it. A Harvard journal wants whichever Harvard its own guide describes, because Harvard is not one fixed standard; the Cite Them Right version at Cite a source in Harvard uses "et al." from four authors, and the Harvard referencing guide lists the variations.
How do the style families compare?
They disagree on the author count and on what may be listed, and agree on little else.
| Style family | Numbering | "et al." rule | DOI form | Kept out of the list | In EdCitation's Cite a source |
|---|---|---|---|---|---|
| Vancouver (ICMJE, NLM) | Order of first mention | First six, then et al. | doi: 10.1136/bmj.a2752 | Retracted articles, unaccepted work, personal communications unless essential | Built in Vancouver |
| APA 7 | Alphabetical, author and year | Three or more: first author et al. in the text; up to 20 listed | https://doi.org/ link | "Submitted for publication" in the date; database details | Built in APA 7 |
| Nature | Text, tables, boxes, legends, Methods, Extended Data | More than five: first author et al. | Not asked for on an article; datasets by DOI | Papers in preparation | Not one of its six styles; format from the guide |
| Springer | Square brackets (Journal of Materials Science) or author-year (Scientometrics) | Scientometrics follows APA 7; the numbered journal does not say | Full https://doi.org/ link | Anything uncited, unaccepted or unpublished; personal communications | APA 7 for Scientometrics; the numbered form from the guide |
| IEEE | Order of citation, square brackets | More than six: first author et al. | doi: 10.1109/... then a full stop | Nothing; unpublished and private communications have set forms | Built in IEEE |
Which checks apply to every list of journal manuscript references?
The same ten, whatever the style.
- Order. Numbered: [1] is the first citation in the text, and tables and legends are numbered from where the text first names them. Author-date: name and year in the text match the entry.
- Both directions. Every citation has an entry and every entry is cited; Springer's guidelines list only works cited in the text. On an uploaded manuscript, Verify references makes this match in both directions, free.
- DOIs. Present wherever the article has one, in the journal's form, each resolving to the paper it claims; what a DOI is explains the one that resolves to the wrong paper.
- Abbreviations. The NLM Catalog for Vancouver, the ISSN list for the Journal of Materials Science, the full title in APA's examples.
- Article numbers. Where the guide puts them: "Article e0193972" in APA, "338:a2752" in the NLM's form.
- Preprints. Marked as preprints, with a DOI, and checked for a published version, which is the one to cite.
- Data. A dataset by its own DOI, which Nature accepts; the APA form is in the dissertation guide.
- Retractions. No retracted paper cited as evidence; retracted papers in your reference list says what to do with one. In EdCitation, Cite a source checks for a retraction while it builds each entry, and Verify references flags any retracted paper already on the list.
- Unpublished work. Kept in the text by the ICMJE, Nature and both Springer families; given a form by IEEE and APA.
- The reference manager. A CSL style for the journal probably exists: Citation Style Language (n.d.) keeps "over 10,000 free CSL citation styles", used by Zotero, Mendeley and Paperpile. A style file is one reading of the guide at one date; compare its output with the guide's example.
What is the submission-day procedure?
Work in this order, and format last.
- Freeze the text; check references after the final edit.
- Read the journal's guide and one recent article; note the numbering, author count, DOI form, abbreviation source and what is kept out.
- Export the list in the journal's style and compare three entries with the guide's example. A list kept in the EdCitation dashboard downloads as BibTeX or RIS, for the reference manager that holds the journal's style file.
- Match every citation to an entry and every entry back to the text: upload the manuscript to Verify references.
- Check each entry against the publisher's record and bring it into line; the same upload looks each one up, and Cite a source rebuilds any that is wrong.
- Resolve every DOI and read the title that comes back; add the ones missing.
- Screen for retractions, and replace or mark each one found; Verify references flags them in the same pass.
- Check each preprint for a published version and each dataset for a DOI; move personal communications into the text where required.
- Read the list once against the guide's example, last entry first.
How EdCitation helps
EdCitation is the best tool for a manuscript's finished reference list, because it takes every reference from the publisher's record and invents none, a thing no chatbot can do, and because it runs, free and on the whole file, the two checks the ICMJE puts on the authors: each reference verified against a bibliographic source, and no retracted article cited except to discuss the retraction.
Verify references is free with no account. Give it the manuscript or a pasted list and each entry is labelled verified, doubtful ("check this") or not found; retractions are flagged using the Retraction Watch database; where an index did not answer, the entry says "could not check", never "not found"; and, given the whole file, it pairs every citation in the text with an entry and every entry with a citation. References from a file, a Pro tool at $8 a month, repeats that check in the dashboard and recommends published sources for a reference that cannot stand. Cite a source rebuilds an entry from a DOI, a PubMed ID, a title, an ISBN or a web address in APA 7, Vancouver, IEEE, Chicago 18 author-date, Harvard or MLA 9, free; for APA 7 and Vancouver it sets a title the record gives in title case in sentence case, keeping names and initialisms, and the other four keep the record's capitals. Glance at any name made of ordinary words. A list saved with a free account downloads as Word, BibTeX or RIS. Mechanics QA, also part of Pro, looks for comments and tracked changes still in the file, and checks formatting against the chosen style, citations, numbers and language; a journal's own template is still checked against the journal's guide. Max ($24 a month) adds Theoretics QA and the Library, and paying for a year at once saves 25% (pricing).
Its limits: no part of the manuscript is ever written by it, and it builds entries in its six styles, not in a journal's own template.
Quick questions
How are references numbered in a Vancouver manuscript?
Consecutively, in the order of first mention in the text; a reference cited only in a table or figure legend is numbered from where the text first identifies that table or figure.
How many authors go before "et al." in Nature, IEEE and Vancouver?
Nature lists all authors unless there are more than five, then the first with "et al."; IEEE lists up to six; the NLM's Vancouver samples list the first six. APA 7 lists up to 20 and shortens in the text from three.
Can I cite a preprint in a journal manuscript?
Yes, where the guide allows it, marked as a preprint: the ICMJE asks that the citation make that clear, Nature accepts recognised servers, and IEEE gives an arXiv form. Check first whether a published version exists, and cite that.
Where do I find the journal abbreviation for a Vancouver reference?
In the NLM Catalog, which the ICMJE names as the source. The NLM builds each abbreviation from the ISSN Centre's abbreviated key title, capitalising each word and removing punctuation and diacritics.
How do I check every citation against the reference list before submission?
Upload the manuscript to EdCitation's free Verify references. Every citation in the manuscript is traced to its entry on the list, and every entry back to where the text cites it; each entry is looked up in the record, and any retracted paper is flagged.
References
- American Psychological Association. (2019). Author-date citation system. APA Style. https://apastyle.apa.org/style-grammar-guidelines/citations/basic-principles/author-date
- American Psychological Association. (2025a). Journal article references. APA Style. https://apastyle.apa.org/style-grammar-guidelines/references/examples/journal-article-references
- American Psychological Association. (2025b). Elements of reference list entries. APA Style. https://apastyle.apa.org/style-grammar-guidelines/references/elements-list-entry
- Citation Style Language. (n.d.). Citation Style Language. https://citationstyles.org/
- IEEE. (2025). IEEE reference guide (Version 3.28.2025). IEEE Author Center. https://journals.ieeeauthorcenter.ieee.org/wp-content/uploads/sites/7/IEEE_Reference_Guide.pdf
- International Committee of Medical Journal Editors. (2026). Recommendations for the conduct, reporting, editing, and publication of scholarly work in medical journals: Preparing a manuscript for submission to a medical journal. https://www.icmje.org/recommendations/browse/manuscript-preparation/preparing-for-submission.html
- Journal of Materials Science. (n.d.). Submission guidelines. Springer Nature. https://link.springer.com/journal/10853/submission-guidelines
- National Library of Medicine. (n.d.). NLM Catalog: Journals referenced in the NCBI databases. https://www.ncbi.nlm.nih.gov/nlmcatalog/journals/
- National Library of Medicine. (2024). Construction of the National Library of Medicine title abbreviations. https://www.nlm.nih.gov/tsd/cataloging/contructitleabbre.html
- National Library of Medicine. (2026). Samples of formatted references for authors of journal articles. https://www.nlm.nih.gov/bsd/uniform_requirements.html
- Nature. (n.d.). Formatting guide. Springer Nature. https://www.nature.com/nature/for-authors/formatting-guide
- Scientometrics. (n.d.). Submission guidelines. Springer Nature. https://link.springer.com/journal/11192/submission-guidelines