Preprint vs. Peer-Reviewed Article: Which Version Should You Cite?

The same research exists twice. Which version should you cite?

You find an excellent preprint and add it to your literature review. Several months later, you discover that the same study has now been published in a peer-reviewed journal.

Should you replace the preprint citation with the journal article? Cite both? Keep the preprint because that was the version you originally read?

In most ordinary situations, the answer is straightforward: cite the peer-reviewed journal article when it is the subsequent formal publication of the same research and it supports the claim you are making. But the principle becomes more complicated when the two versions contain different data, analyses, outcomes, methodological details, or conclusions.

The key is to stop thinking of the preprint and journal article as two independent pieces of evidence. They are usually two reports or versions of the same underlying study, and your citation should point to the version that actually supports the information you are using.

Usually cite the peer-reviewed article, but first check what changed

Cite the report that actually supports the claim

A preprint and journal article are usually not two separate studies

The most important conceptual distinction is between a study and a report of a study.

A research project can generate several reports. The same underlying study might appear as a registry record, conference abstract, dissertation, preprint, journal article, supplementary appendix, secondary analysis, or follow-up publication.

Cochrane guidance explicitly treats the study rather than each individual report as the unit of interest in a systematic review. Multiple reports should be linked because treating them as independent studies can duplicate participants and introduce bias.

A preprint and its subsequent journal article therefore usually provide two versions of evidence about one study, not two independent confirmations of the same result.

Why is the peer-reviewed article usually preferred?

The journal article is generally the later formal publication. Between preprint posting and journal publication, the manuscript may undergo editorial assessment, external peer review, revision, additional analysis, correction, clarification, and journal production.

That does not mean peer review guarantees that the journal article is correct. It does mean the published version has passed through an additional scholarly process that the preprint had not completed when it was posted.

The journal version also generally provides the stable bibliographic information expected in scholarly citation: journal title, volume or article number, publication date, and DOI associated with the formal publication.

For a claim that appears unchanged in both versions, citing the later peer-reviewed article is therefore usually the clearest choice.

Do not replace the citation without checking that the paper is actually the same study

Finding a journal article by the same authors on the same topic is not sufficient to establish that it corresponds to the preprint.

Compare identifying characteristics such as:

• study registration number;
• authors;
• study population and setting;
• sample size and baseline characteristics;
• recruitment period;
• intervention or exposure;
• comparison groups;
• outcomes;
• follow-up period;
• study acronym or project identifiers.

The title may have changed substantially. The sample size may also have increased if recruitment or follow-up continued after the preprint was posted.

Cochrane recommends using multiple study characteristics when linking reports and contacting investigators when uncertainty cannot otherwise be resolved.

Do not assume the journal article contains exactly the same information

Empirical comparisons show that many preprints and their later journal publications remain broadly similar, but they are not necessarily identical.

One study comparing 547 medRxiv preprint-journal article pairs found concordant primary endpoints in 97.6% of pairs and concordant overall study interpretations in 96.2%. However, exact numerical primary-endpoint results were concordant in 81.1% of pairs. Sample sizes also differed in some publications.

Other research comparing preprints and journal articles has documented changes in reported outcomes, numerical results, statistical significance, analyses, follow-up periods, and interpretation.

The practical implication is simple: publication usually strengthens the case for citing the journal article, but it does not justify assuming that every statement supported by the preprint is also supported identically by the published version.

What kinds of differences should you look for?

You do not necessarily need to conduct a forensic comparison of every sentence. Focus on information relevant to the claim you intend to cite.

The more central the study is to your argument, the more worthwhile this comparison becomes.

What if the information you need appears in both versions?

If the relevant finding, method, or conclusion appears consistently in both documents, cite the peer-reviewed article.

There is normally little value in citing both versions merely to show that you found both. Doing so can make one study look like two independent sources to readers who do not inspect the references closely.

For example, suppose the preprint and journal article both report essentially the same primary finding. If you write a sentence describing that finding, the journal article will usually be sufficient.

The preprint remains part of the study's publication history, but that does not mean it needs to appear beside every citation to the final paper.

What if the preprint contains information missing from the journal article?

This is where an automatic "always cite the published paper" rule breaks down.

A preprint may contain methodological details, secondary outcomes, sensitivity analyses, supplementary explanations, tables, figures, or other material that does not appear in the journal article. Journal word limits and editorial decisions can result in material being shortened or removed during publication.

If you make a claim that is supported only by the preprint, cite the preprint for that claim and make its status clear.

Do not cite the journal article merely because it is the later publication if the information you are attributing to the study cannot actually be found there. A citation is evidence for a statement, not a ceremonial preference for the more prestigious document.

When should you cite both versions?

Citing both is appropriate when the existence of both versions is itself relevant to your point.

For example, cite both when:

• you are discussing how the study changed between preprint and publication;
• a relevant result appears only in the preprint while other information comes from the journal article;
• you are documenting changes in sample size, methods, analyses, or outcomes;
• the authors substantially changed their interpretation;
• you are conducting methodological research on preprint-to-publication changes;
• both reports provide complementary information needed to describe or appraise the study.

In these cases, the dual citation tells the reader something meaningful rather than merely duplicating references.

What if the preprint and published paper report different results?

Do not resolve a discrepancy merely by deleting the preprint from your notes and keeping the journal article.

First establish what changed. Was the sample larger? Was follow-up extended? Did the authors correct an error? Was the outcome definition altered? Did peer review prompt a different statistical analysis? Are you comparing an interim analysis with a final analysis?

The later peer-reviewed article will usually be the appropriate source for the final reported result, but the earlier result can remain methodologically relevant, especially when the difference affects interpretation of the study.

When the preprint and published paper report different results, investigate and document the discrepancy rather than treating publication status as an explanation by itself.

Do not assume every change was caused by peer review

A preprint may change before journal publication for many reasons. Authors may discover an error independently, receive informal feedback, collect additional data, refine an analysis, respond to editorial requests, or make changes requested by peer reviewers.

Unless the editorial history or authors provide that information, you usually cannot know which changes resulted specifically from peer review.

It is therefore safer to say that a finding changed between the preprint and journal publication than to claim that "peer review corrected" it without evidence.

The published article is not automatically more complete

Peer-reviewed journal articles often improve reporting. A study comparing biomedical preprints with published articles found slightly higher reporting quality, on average, in peer-reviewed versions.

But later does not necessarily mean longer or more comprehensive in every respect. Journal formatting and word limits can reduce details available in an earlier manuscript. Supplemental files may restore some of that information, but not always.

This is another reason Cochrane advises reviewers not to discard secondary reports of the same study. Different reports can contain additional outcomes or useful information about study design and conduct.

For systematic reviews, collate rather than choose too early

If you are conducting a systematic review, the immediate task after finding both versions is not necessarily to throw one away. Link them as reports of the same study.

Cochrane requires multiple reports of a study to be collated so that the study, rather than the individual publication, remains the unit of interest. Reviewers should choose and justify which report serves as the primary source for study results, especially when reports conflict, while retaining secondary reports because they may contain valuable additional information.

This has several practical consequences:

• do not count the preprint and article as separate studies;
• extract complementary information across reports where appropriate;
• identify which report supplied each important result;
• resolve discrepancies rather than choosing whichever value is more convenient;
• update risk-of-bias or quality assessments when later reporting provides additional information.

This approach is more defensible than assuming that publication turns every earlier report into irrelevant clutter.

If your review originally included the preprint, update more than the reference

Suppose you completed data extraction from a preprint and then discover the journal article during an updated search.

Do not simply change the bibliography.

Recheck:

• eligibility;
• sample size;
• participant characteristics;
• outcome definitions;
• numerical results;
• analysis methods;
• risk-of-bias information;
• funding and conflicts;
• study conclusions.

If any of these changed, your synthesis may require updating. A publication-status update can therefore become a data update.

What if you cited the preprint before the journal article existed?

You did not necessarily do anything wrong. If the preprint was the appropriate available source at the time and your context allowed preprint citation, it supported the claim based on the evidence then available.

For a manuscript that has not yet been finalized, however, check cited preprints again before submission or publication. If a peer-reviewed version has appeared, update the citation where appropriate and verify that the claim remains supported.

This is one reason researchers who include preprints in a literature review should record their publication status and revisit important preprints during the final search update.

What if the journal article has a different title?

This is common enough that title matching alone is unreliable.

Preprint platforms sometimes link directly to subsequent journal publications. When they do not, search the authors, study topic, registration number, study acronym, and distinctive methodological characteristics.

If the titles differ but the study identifiers, participants, methods, and other characteristics match, treat the documents as related reports rather than assuming they represent independent studies.

What if the preprint has several versions?

A preprint server may preserve multiple revisions. If you cite a preprint because no journal article exists, use the current version unless you have a specific reason to discuss an earlier one.

If your argument concerns how the research changed over time, version information becomes part of the evidence. A first preprint version, revised preprint, and journal article may document three stages of the same study's reporting history.

Be precise about which version supports which statement.

What if the journal article is behind a paywall but the preprint is freely available?

Accessibility alone does not normally make the preprint the preferable citation when a peer-reviewed article exists and supports the claim.

The citation identifies the source of the scholarly claim, not simply the easiest copy for readers to obtain. Depending on publisher policy, the preprint may remain useful as an openly accessible version, but it should not be represented as the peer-reviewed article.

If you discuss material unique to the preprint, cite the preprint. If you discuss findings established in the formal publication, cite the article.

Retractions, corrections, and withdrawal notices override simple version preferences

"Always cite the latest version" also needs qualification.

If the journal article has been retracted, corrected, or otherwise formally amended, investigate the current publication record before citing it as ordinary evidence. Likewise, if the preprint has been withdrawn, examine the withdrawal notice.

A preprint should not be used to circumvent a later retraction of substantially the same research, and a retracted article should not be cited as ordinary valid evidence simply because it is the peer-reviewed version.

Publication history is part of evidence appraisal, not merely bibliography management.

How to choose the right citation when both versions exist

Common mistakes when choosing between a preprint and published article

Choose the citation according to the evidence behind your sentence

The most useful rule is not "published always beats preprint." It is: identify the study, determine which report supports your statement, and prefer the later peer-reviewed report when the relevant evidence is adequately represented there.

For a systematic review, preserve all relevant reports in the study record even when one becomes the primary source for results. For ordinary scholarly writing, you usually do not need to display that entire documentary history unless it matters to the argument.

The citation should make the evidence easier to verify, not make the bibliography look more populous than the underlying research actually is.

Before replacing a preprint with a journal citation

Frequently asked questions about citing preprints and published articles

Prefer the published article, but cite the evidence you actually used

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.