Abstract

An abstract is a standalone summary placed at the start of a research paper, stating the study’s purpose or background, methods, principal results, and conclusions. Its function is operational, not decorative: a reader scanning search results, or a database indexing the paper, uses the abstract alone to judge whether the full text is relevant enough to read. Because of that function, an abstract has to be intelligible on its own — no citations by reference number, no abbreviations that are only defined later in the body text, and no assumption that the reader has seen (or will go on to read) any other part of the manuscript.

Structured vs. unstructured abstracts

Journals require one of two formats. A structured abstract uses explicit labeled subheadings — commonly Background, Methods, Results, and Conclusions, though biomedical journals following a more elaborate clinical-research format may add headings such as Design, Setting, Participants, or Main Outcome Measures. An unstructured abstract covers the same ground (purpose, approach, findings, significance) but as a single continuous paragraph with no subheadings. ICMJE’s Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work specify that original research, systematic reviews, and meta-analyses require a structured abstract stating the study’s context or background, purpose, basic procedures, main findings (with effect sizes and their significance where possible), and principal conclusions — but ICMJE explicitly leaves the exact subheadings and word count to the individual journal, noting that “the format required for structured abstracts differs from journal to journal.” By contrast, PLOS ONE’s submission guidelines call for an unstructured abstract: a single paragraph describing the study’s main objective, how it was done (without excessive methodological detail), and the most important results and their significance.

How long should an abstract be?

There is no single universal word limit — ICMJE does not prescribe one, only noting that a separate word count for the abstract is useful for editorial assessment, and instructs authors to follow the target journal’s own instructions. In practice, journal-set limits commonly fall somewhere between roughly 150 and 350 words. PLOS ONE, for example, caps its unstructured abstract at 300 words. A structured abstract’s per-section space is typically tighter still, since the word budget is split across several labeled fields (a sentence or two of background, a couple of sentences on methods, two or three on results, and a short conclusions statement) rather than left to a single free-flowing paragraph. Always confirm the exact limit and required format in the specific journal’s (or conference’s) instructions for authors before submission — it is not a fixed, portable number across the literature.

What an abstract is not

Three genuine mix-ups are worth ruling out explicitly:

• Not a table of contents. A table of contents lists a document’s section headings (and often page numbers) so a reader can navigate it — it doesn’t state what the study found or why it matters, and it isn’t written to stand alone.
• Not an introduction. The introduction is part of the full text: it situates the study within prior literature and sets out the research question or hypothesis, but it is written to lead into the methods and results that follow, not to summarize them. An abstract has to cover the whole paper, including results and conclusions the introduction hasn’t reached yet; the introduction doesn’t, and isn’t indexed as a standalone surrogate for the paper the way an abstract is.
• Not a keyword list. Keywords (or index terms) are short topic tags attached to a paper to aid search and subject indexing. They help a database classify the paper; they don’t summarize its purpose, methods, or findings the way an abstract’s prose does.

Related concepts

A conference abstract is a distinct, closely related genre: a short summary submitted to a conference for presentation selection, published in a conference abstract book or supplement, and typically lacking the full peer review a journal article’s abstract accompanies. Preprints and working papers also carry their own abstracts, following the same standalone-summary function even before formal peer review. For the review process an abstract’s parent manuscript goes through after submission, see CASRAI’s peer review guide. For CASRAI’s broader scholarly-writing coverage, see the Scholarly Writing & Publishing hub.