Citations in the abstract: why journals say no and what to write instead.

In most journals, no. PLOS ONE's submission guidelines state that abstracts "should not include citations," and Nature's formatting guide asks for a summary of roughly 150 words, unreferenced. The abstract is meant to stand alone as a complete miniature of the paper, so a sentence that leans on someone else's work has to carry it in words, not in a reference marker.

What does the no-citations rule actually say?

The PLOS ONE version is one line in the submission guidelines: "Abstracts should not include citations." Nothing about exceptions, nothing about limits on how many. The count allowed is zero.

That line sits beside the rule capping the abstract at 300 words, and the two work as a pair. One limits how much you can say, the other limits what you can lean on while saying it. Nature compresses the same instruction into a single phrase, asking for abstracts of about 150 words, unreferenced. Different wording, same demand. Whatever the abstract claims, it claims on its own authority. Abstract length varies far more between journals than this rule does; while the word caps range from 150 to 300, the expectation that the abstract stands free of the reference list barely moves.

Why do journals ban citations in the abstract?

Because the abstract travels without the paper, and a citation is a pointer into a document the abstract's reader does not have.

Your abstract gets read in places where your reference list does not exist: a database record, a search result page, an email alert, the screening pile of someone else's systematic review. In numbered reference styles the failure is easy to picture. Vancouver-style journals cite by a bracketed number that resolves to an entry in a list at the end of the manuscript. Strip the list away, which is exactly how a database displays your abstract, and the number points at nothing.

Author-date citations fail more politely, but they still fail. A surname and a year in parentheses tell the reader that the evidence lives somewhere else, which is the move the format forbids. The abstract reports what your study found, in sentences that survive on their own.

The word budget argues the same way. At 150 or 300 words, every phrase has to earn its place, and a citation spends several of them announcing that support exists instead of saying anything.

How do you rewrite a sentence that leans on a citation?

Three patterns cover nearly every case: state the fact as established knowledge, name the source in prose, or reframe the sentence around your own result.

The first pattern handles background claims. If the cited fact is settled in your field, assert it plainly and let the introduction carry the reference. "Prior work has shown that sleep quality declines with age (Smith et al., 2019)" becomes "Sleep quality declines with age." The abstract states what the field already accepts. The formal citation waits for the opening paragraph of the introduction, where the reference list can back it.

The second is for when the source's identity matters. When the reader needs to know which instrument, dataset, or trial you mean, name it in words. "We measured symptoms using a validated questionnaire (Beck et al., 1996)" becomes "We measured symptoms with the Beck Depression Inventory." The name does the citation's work and the sentence still stands alone. Cohorts follow the same move; "UK Biobank" identifies a dataset more precisely than any reference number could.

The third covers comparison claims. "Our results are consistent with previous findings (Chen et al., 2021)" asks the reader to know a paper they cannot see. Restate the comparison as your own finding: "The effect held in both cohorts, at roughly half the size reported in earlier trials." The comparison survives, the pointer is gone, and the detailed reckoning with prior work moves to the discussion, where it belonged anyway.

None of these rewrites weaken the claim. A citation in an abstract only ever promised evidence. Stating the finding outright makes the same promise in fewer words.

What if your whole study responds to one earlier paper?

Name that paper in words. A replication, a comment, or a reanalysis takes an earlier study as its subject, and hiding that fact from the abstract would misdescribe your work.

Identify the study the way you would in conversation: authors or venue, year, and the claim at stake. A sentence like "We report a preregistered replication of a 2016 study linking sleep loss to next-day risk taking" tells the reader precisely what the paper is about, with no marker to decode. If the original is famous enough, the authors' names alone can do it. The formal citation then appears at the first mention in the introduction.

The rule holds even here. The abstract stays self-contained because the sentence itself carries the identifying information; no list is needed to make sense of it. What changes is only how much of your word budget the earlier study deserves, and for a replication the honest answer is a full sentence.

Who catches an abstract citation, and when?

The journal's technical check, usually before an editor reads the science.

Editorial staff screen new submissions against the format rules, and a reference marker in the abstract is one of the easiest violations to spot, because the abstract is pasted into the submission system as bare text. The manuscript comes back for correction, and the round trip costs days. Format slips also rarely arrive alone; they tend to travel together with the other problems that precede a desk rejection.

Run the check yourself before the journal does. When Waraq reviews a manuscript against PLOS ONE's rule set, a citation in the abstract gets flagged with the journal's own line, "Abstracts should not include citations," quoted in the margin beside the sentence that breaks it, so the fix happens before any editor sees the file.

Write the abstract last, from the finished paper, and write it as if the reference list did not exist. For the abstract's most important readers, it never does.