A preprint is a complete manuscript posted to a public server before or during peer review, and citing one correctly means doing two things a journal-article citation doesn’t require: naming the preprint server as the source of record (not the journal you hope it ends up in), and marking clearly that the version you’re pointing to has not been peer-reviewed. Get either wrong and the citation misrepresents the evidentiary status of the work it points to.
The elements a preprint citation needs that an article citation doesn’t
Every major style guide asks for the same underlying information, just arranged differently. A complete preprint citation needs:
- Preprint server name as the publisher/source — bioRxiv, medRxiv, arXiv, SSRN, PsyArXiv, OSF Preprints, and similar servers stand in the position a journal name would occupy, but they are not journals and shouldn’t be formatted as one.
- An explicit non-peer-review marker — either a bracketed tag like [Preprint], a repository label that already signals it (e.g., “bioRxiv”), or, in narrative text, an explicit statement that the work has not completed peer review.
- The preprint’s own persistent identifier, usually a DOI. Most preprint servers now mint a DOI at posting through DataCite or Crossref, distinct from any DOI the eventual journal article receives.
- A version or posting date, since preprints are frequently revised. Most servers version each revision (v1, v2, v3…); citing without a version can point a reader to a different revision than the one you actually used.
- No volume, issue, or page numbers — preprints don’t have them, because they haven’t gone through a journal’s production process.
These are the same underlying concerns — persistence, versioning, unique identification — that drive data citation practice under the FORCE11 Joint Declaration of Data Citation Principles. A preprint, like a dataset, is a versionable research object rather than a fixed text, so citation conventions for both have converged on the same fix: name the persistent identifier and the version, don’t rely on a static reference the way you would for a bound journal issue.
APA style (7th edition)
APA’s official guidance treats a preprint as its own reference type, structured as:
Author, A. A. (Year). Title of preprint. Repository Name. https://doi.org/xxxxx
APA’s own published example:
Hampton, S., Rabagliati, H., Sorace, A., & Fletcher-Watson, S. (2017). Autism and bilingualism: A qualitative interview study of parents’ perspectives and experiences. PsyArXiv. https://doi.org/10.31234/osf.io/76xfs
Name the repository (PsyArXiv, bioRxiv, arXiv) as the source in place of a journal name, and omit volume/issue/page elements entirely, since a preprint doesn’t have them. APA’s guidance is explicit on the underlying preference: where possible, cite and use the final published version of a work rather than the preprint; cite the preprint version specifically only when that’s the version you actually used, or when no published version exists yet.
MLA and Chicago style
Neither MLA (9th edition) nor Chicago maintains a dedicated preprint template the way APA now does; both apply their general web-source/container logic instead:
- MLA: Author. “Title of Preprint.” Repository Name, Date posted, DOI or URL.
- Chicago (author-date): Author. Year. “Title of Preprint.” Preprint, Repository Name. DOI or URL.
In both cases, the repository stands in the position a journal or publisher would normally occupy, and the word “Preprint” (or an equivalent bracketed note) should appear somewhere in the citation to flag the peer-review status, since neither style’s general web-source template does that automatically. As with data citations, style guides revise exact punctuation between editions — check the current edition you’re required to use rather than a remembered template, and note that many preprint servers auto-generate a suggested citation in several formats directly on the preprint’s landing page.
Vancouver / NLM style (biomedical and clinical fields)
Vancouver-style citation, based on NLM’s Citing Medicine conventions, is the format most commonly required in medical and clinical journals, and it makes the non-peer-reviewed status of a preprint an explicit, bracketed part of the citation itself rather than something left to context:
Author(s). Title of preprint. Repository name [Preprint]. Year [posted/revised date; cited date]. Available from: URL or DOI
A representative example, closely following published NLM-style templates:
Riad R, Denais M, de Gennes M, et al. Automated speech analysis for risk detection of depression, anxiety, insomnia, and fatigue. medRxiv [Preprint]. 2024 [posted 2024 Sep 23; cited 2024 Sep 24]. Available from: https://doi.org/10.1101/2024.03.20.24304577
The bracketed [Preprint] tag is doing real work here, not decorative: Vancouver-style citations are heavily used in clinical contexts where a reader needs to immediately register that a claim rests on unreviewed evidence, and the format is built to surface that at a glance rather than requiring the reader to check the source name.
What ICMJE actually says about preprints and citing them
The International Committee of Medical Journal Editors (ICMJE) Recommendations address preprints directly, and the guidance is relevant to research administrators well beyond the citation-formatting question:
- Authors should inform a journal if a submitted manuscript has already been posted as a preprint, and provide a link to it.
- It is the author’s responsibility — not the journal’s — to make sure the preprint is amended to point readers to any subsequent, formally published version.
- When a preprint is cited in a submitted manuscript or a published article, “the citation should clearly indicate that the reference is a preprint” — this is the same discipline the style-guide formats above are built to enforce mechanically.
- Where a preprint has since been published in a peer-reviewed journal, authors should cite the published version rather than the preprint “whenever appropriate.”
- ICMJE specifically cautions authors to be careful about citing preprints that were posted and never subsequently published in a peer-reviewed journal — the absence of a published version is itself evidence worth weighing before relying on the claim.
ASAPbio, the scientist-led initiative that produces much of the field’s best-practice guidance on preprint posting, screening, and versioning, takes a consistent position: preprints are increasingly citable, community-accepted research objects, but the citation itself needs to carry the non-peer-review flag so downstream readers, funders, and assessment processes aren’t misled about what kind of evidence they’re looking at.
A practical checklist before you submit a preprint citation
- Check for a published version first. Search the preprint server’s own landing page (most now display a “Published in…” banner and link once a journal version exists) or search the title directly. If a peer-reviewed version exists, cite that one instead, per ICMJE and most style guides’ stated preference — unless you specifically need to reference claims, data, or wording unique to the preprint version.
- Confirm you’re citing the version you actually used. If the preprint has multiple versions (v1, v2, v3), the citation’s DOI or URL should resolve to the specific version, not just the preprint’s general landing page, if the server supports version-specific links.
- Include the non-peer-review marker, whether that’s a style guide’s bracketed [Preprint] tag, the repository name doing that work implicitly (as in APA), or an explicit sentence in narrative text.
- Use the preprint’s own DOI, not a placeholder or the eventual journal DOI you’re anticipating — those are different, separately registered identifiers even when the underlying manuscript is the same.
- Follow your target journal’s specific instructions. Some journals restrict or specifically govern how preprints can be cited or discussed within a submitted manuscript; check author guidelines rather than assuming a general style-guide template is sufficient on its own.
Frequently asked questions
Do I need to cite a preprint differently if it has since been published?
Yes — once a peer-reviewed version of record exists, cite that version rather than the preprint, per ICMJE’s stated preference and most style guides’ general guidance. Cite the preprint specifically only when you’re referencing something unique to that version, or no published version exists yet.
Does a preprint have its own DOI, separate from the eventual journal article?
Yes. Most preprint servers register a DOI for the preprint itself at posting, through DataCite or Crossref. If the manuscript is later published in a journal, the journal article typically receives its own, separate DOI — the two identifiers point to two distinct, individually citable objects.
Is it acceptable to rely on a claim that only exists in a preprint that was never published?
It can be, but with caution. ICMJE specifically flags this scenario as one to be careful about, since the absence of a subsequent peer-reviewed publication may reflect unresolved issues raised in review, not just a slow or abandoned submission process. Weigh it accordingly and flag the non-peer-reviewed status clearly wherever the claim is used.
How do I know which citation style to use for a preprint?
Use whatever style your target journal, institution, or funder requires for the manuscript as a whole — there’s no separate “preprint style” independent of the base citation style (APA, MLA, Chicago, Vancouver, or a journal-specific variant). What changes for a preprint, within any of those styles, is which elements you supply (repository name instead of journal, no volume/issue/pages, an explicit non-peer-review marker) not the overall style family.
For more on how preprint servers themselves operate — screening practices, versioning, and how to choose one as an author or research office — see How to Choose a Preprint Server. For the underlying identifier mechanics referenced throughout this guide, see How to Find, Cite, and Use a DOI and the Scholarly Publishing pillar page.
