Can You Cite a Preprint in a Literature Review? (2026)

Preprint: a complete draft of a research paper that its authors have posted publicly before journal peer review. The United States National Library of Medicine defines preprints as "complete and public drafts of scientific documents, not yet certified by peer review", and warns that "any aspect of the research, including the results and conclusions, may change as a result of peer review" (PMC, NIH Preprint Pilot).

What is the difference between a preprint, an accepted manuscript and the version of record?

A preprint has had no peer review, an accepted manuscript has passed peer review but has not been formatted by the publisher, and the version of record is the final published article with its journal citation. The table sets the three side by side, because the version you read decides the reference you write.

Sources: bioRxiv, About for the preprint row, and De Montfort University Library for the post print row, both read in October 2026.

Each version also carries its own DOI. Crossref asks preprint publishers to give every new version of a preprint a new DOI, linked to the others through a relationship of type isVersionOf, and to link the preprint to its accepted manuscript or version of record once one exists (Crossref, Introduction to posted content). The preprint and the journal article are therefore two records with two DOIs, which is the machine readable reason a reference manager sometimes holds two entries for what looks like one paper.

De Montfort University’s referencing guidance draws the same line for students. A post print is referenced like the journal article, while for a preprint the reference has to make it clear "that this is a preprint and not the final approved article" (De Montfort University Library).

Will my examiner accept a preprint as a source in a literature review?

There is no universal rule, and that is the honest answer: acceptance depends on your discipline and your department, not on a single global policy. University guidance that does exist leans towards permitting preprints with a visible label. The University of Surrey states that "as long as the preprint versions are clearly indicated as non-peer-reviewed, it is the responsibility of the readers to evaluate the research" (University of Surrey, Preprints guidance).

The second condition comes from the same direction. A December 2025 guide by Sharjeel Chaudhry and Zarmina Ehtesham on Springer Nature’s Research Communities site tells researchers to "Check for a peer-reviewed version. If the paper is later published in a journal, always cite the final version instead of the preprint" (Citing a Preprint, Research Communities by Springer Nature).

In practice an examiner reads one labelled preprint supporting a claim that also has peer reviewed support very differently from a chapter resting on unreviewed papers. Ask your supervisor before a section leans on preprints, and where a departmental rule exists, follow the rule.

How often do a preprint’s results change before it is published?

Often enough that the check matters, and there is a measured figure for it. Sommer and colleagues followed 329 prevention research preprints posted to medRxiv: 48.9% were published in a peer reviewed journal within 12 months, with a median of 5.3 months from upload to publication. Among those published, the magnitude of the primary outcome effect estimate changed in 16.8%, with a major change of more than 25% in 4.4%. The style or wording of the conclusion changed in 42.2%, and its content in 3.1% (Sommer et al., 2023, Scientific Reports 13, 17034).

Sommer and colleagues are blunt about what that means: "The reliability of preprint articles for evidence-based decision-making is questionable." Read the figures for what they are, though. The analysis covers prevention research, not every field, and the content of most conclusions survived peer review unchanged. What the analysis rules out is the assumption that a preprint and its published version say the same thing, so the number you quote in a literature review should come from the version you actually read.

How do I check whether a preprint has since been published in a journal?

Start at the preprint’s own landing page, because as of October 2026 the servers do most of this work for you. bioRxiv "will usually automatically add a link to the published version within several weeks of journal publication", and asks readers to wait two to three weeks before contacting staff if that link has not appeared (bioRxiv, About). Crossref requires the preprint publisher to link the preprint to its accepted manuscript or version of record within seven days of a match alert, with the link shown where a reader sees it without scrolling (Crossref).

Those matches run on the title and the first author, and bioRxiv notes that a match is sometimes missed "because the title or authors have changed". So a missing link is not proof that no journal version exists. Three checks catch almost everything. Open the DOI and see whether the landing page announces a published version. Search the exact title in a scholarly index, since a changed title is the usual reason an automatic match fails. Then search the first author’s surname with two distinctive words from the title, which finds papers whose title and author order both changed.

Litrevu’s free reference checker does something narrower, and it is worth being clear about the limit. The reference checker takes a pasted reference list, tests each entry against Crossref, OpenAlex and doi.org, and flags retractions. It does not report whether a preprint has since been published in a journal, so that check stays manual, one preprint at a time. For retractions across a finished list, our guide on how to check whether a paper you cited has been retracted covers the same sources.

How do I reference a preprint in APA and Harvard?

Both styles keep the author, year and title you already know and then point at the version you read. APA 7th names the repository and gives the preprint’s DOI, and Cite Them Right Harvard adds an explicit [preprint] label. Augusta University’s APA guide also warns that the DOI must take you to "the article you found, e.g. the preprint, not to the formally published version" (Augusta University, APA 7th Citation Guide).

Templates copied from Augusta University’s APA 7th guide and De Montfort University’s Cite Them Right Harvard guidance, read in October 2026. Harvard is a family of house styles rather than one standard, so check your own institution’s version before copying the second row.

The common mistake is a hybrid: the preprint’s text with the journal’s reference. Side by side, with placeholder elements:

• Incorrect, if you read the preprint: Author, A. A. (Year). Title of article. Journal Name, Volume(Issue), pages. https://doi.org/[journal DOI]
• Correct, if you read the preprint: Author, A. A. (Year). Title of article. Repository Name. https://doi.org/[preprint DOI]

The first entry sends your reader to a version you never opened, and if a number moved in review, the number in your sentence will not be on the page they land on. Yes, this is pedantry. It is also the check an examiner can run in under a minute.

Some guides add [Preprint] after the title in APA as well, and the Research Communities guide quoted above puts the reason in one line: "Always include the word ‘Preprint’ in your citation so readers understand the work is not peer-reviewed." If your department’s APA guide shows the label, use it. Element order is otherwise the ordinary one for each style, set out in our APA citation format guide and Harvard referencing guide, and the rules for a source you only met inside another paper are in how to cite a source cited in another source.

What should the sentence in my literature review actually say?

Name the preprint status in the prose, not only in the reference list, because a reader who skims your paragraph never reaches your bibliography. ICMJE’s requirement that a citation "clearly indicate that the reference is a preprint" is satisfied in the reference list, and good practice in a literature review is to satisfy it twice, once where the claim appears.

Two shapes carry it without sounding defensive. "In a preprint posted to medRxiv, (Author, Year) report a similar pattern, though the paper has not yet been peer reviewed." Or, where a preprint sits beside reviewed work: "Two peer reviewed studies (Author, Year; Author, Year) and one preprint (Author, Year) point the same way, and the preprint is the weakest of the three." Those bracketed tokens are placeholders for your own sources, not real citations.

Avoid quoting a preprint’s abstract as though it were a settled finding. The reason is measurable: in Sommer et al., 2023, 4.4% of published prevention preprints had a primary effect estimate that moved by more than 25% between the preprint and the journal version. A hedged sentence costs nothing in an examiner’s eyes.

Should I search preprint servers at all when building a literature review?

Search them if your field posts there and your review is meant to be current, and record that you did. The University of Stirling’s systematic review guidance states plainly that "You may wish to include searches for preprints in your systematic review", listing preprint servers alongside other grey literature sources (University of Stirling, Systematic Reviews: Grey Literature). bioRxiv carries the life sciences and medRxiv the health sciences, and both servers’ own cautions belong in your methods note. medRxiv tells readers that "Preprints are preliminary reports of work that have not been certified by peer review. They should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information" (medRxiv, About), and bioRxiv states that "No endorsement of an article’s methods, assumptions, conclusions, or scientific quality by bioRxiv or openRxiv is implied by its appearance in bioRxiv" (bioRxiv, About).

If your review has an inclusion and exclusion table, put the decision in it: preprints included and labelled, or excluded with a reason. Our guide to grey literature in a literature review covers the same decision for theses, reports and conference papers, and how recent your sources need to be covers the related question of what a currency claim actually requires.

What happens to a preprint I upload to Litrevu?

Litrevu is an AI literature review assistant that turns the papers a researcher has already gathered into a cited first draft, with every citation traceable to the uploaded source. Litrevu drafts only from the papers in your library and reads the PDF you uploaded, so if a journal version of that preprint appeared last month and is not in your library, the draft will not know. Checking for the published version, and naming the preprint status in your prose, both stay your job. The free tier covers 2,000 words with no credit card, as of October 2026, which is enough to see what the drafting step does with a mixed set of reviewed papers and preprints.

Here is the shape of a drafted paragraph, as a format illustration rather than a real result. The names, years and findings are placeholder tokens, not sources:

Early work on this question has appeared mainly outside the journal literature. (Author, Year) report the pattern in a preprint posted to medRxiv, and (Author, Year), in a peer reviewed cohort study, describe a comparable association. The preprint has not been peer reviewed, so the claim rests mainly on the reviewed study.

Every line that comes back is yours to verify against the PDF it came from, rewrite in your own voice and defend in a viva.

Common questions about citing preprints