How do I cite a preprint?

The important part of a preprint citation is not the format, it is the visibility of the status. A reader deciding how much weight to place on your claim needs to know that the supporting work has not been through review, and a reference list entry that looks like a journal article hides that.

So mark it in two places. In the reference, using your style's preprint designation and the server's permanent identifier. And in the text, where the claim carries weight, with a short phrase such as "in a preprint reporting" or "unreviewed work suggests". That phrasing costs three words and does the entire job.

The category is broader than most students assume. Bourne, Polka, Vale, and Kiley define a preprint as a complete description of research that has not completed the usual journal-publication process, and their examples include manuscripts awaiting or undergoing review, commentaries, negative-result reports, datasets with descriptions, and papers already peer reviewed but not yet formally published (Bourne et al., 2017, p. 1). If the item you are citing falls into any of those, label it as a preprint and preserve its persistent public location, the repository record, instead of implying a journal status it does not have.

The identifier matters because preprints are versioned. Servers assign a version number, and the text you read may not be the text a reader finds. Citing the specific version you used is the practice most style guides now recommend and the one that makes your citation reproducible.

The judgment about whether to rely on a preprint is separate from the citation mechanics. Recent, fast-moving fields make preprints unavoidable and citing them is standard. A claim your whole argument depends on is a different matter, and there the right response is caution rather than omission.

Should I cite a preprint at all?

Yes, with the status made visible in your text as well as your reference list. Preprints are legitimate sources, especially for recent work, and the obligation is transparency rather than avoidance. Do not build a load-bearing claim on an unreviewed result without saying that is what you are doing.

Read the preprint more carefully than you would a reviewed paper, precisely because nobody else has. The checks a reviewer would perform, on the analysis, the sample, and whether the conclusions follow, fall to you.

Penders argues that including a source is not enough on its own, and that you should cite it at the point where its specific contribution is used, sometimes citing the same study more than once for different parts of an argument (Penders, 2018, pp. 2-3). Penders names the failure mode "neutral citation", a reference that appears without any statement of why it is relevant. For a preprint that matters more than usual, because the sentence carrying the citation is also the sentence where you say what the work is and what you are relying on, its method, its data, or its preliminary finding.

Bourne and colleagues hold both positions at once. They write that peer review can add substantial value by identifying errors and areas for improvement, and they also treat the preprint as a public, citable disclosure (Bourne et al., 2017, p. 3). The missing review does not make the work worthless, it makes the status a fact your reader needs.

Where a preprint contradicts reviewed work, say so and treat it as an open question rather than a settled reversal. That is an accurate description of the evidence and it protects you if the preprint does not survive review.

What do I do when the preprint is later published?

Switch the citation to the published version and check that the claim you cited survived review, since content changes between the two more often than people assume. Check every preprint citation once before submitting a thesis, because some will have been published since you wrote the chapter.

Set an alert on the preprints you rely on most, or search the authors' names once a year. This takes minutes and prevents the awkward case where a thesis cites an unreviewed version of a paper that has been in print for a year.

The two sources differ in emphasis here. Penders recommends prioritizing reviewed papers over unreviewed or prereviewed sources, and choosing your references again for each publication and separately for each section (Penders, 2018, p. 4). Bourne and colleagues stress that preprints create a public, time-stamped record of priority and can and should be fairly cited, pointing to arXiv's long history in physics (Bourne et al., 2017, pp. 2-3). Both apply. The published version is usually the better citation for a settled finding, and the preprint stays correct when you are discussing its earlier release, its original wording, its priority, or material that appears only in that version.

If the published version differs materially from what you cited, that difference is worth a sentence. Reviewers changed something, and what they changed is information about how solid the original claim was.