How long should an abstract be? What the rules say
APA sets the ceiling at 250 words, and most student abstracts land between 150 and 250. Your own guidelines beat that number, and they vary more than you would think. Louisville caps a thesis abstract at 150 words and a dissertation at 350. Arizona State says 350 for everybody. Whatever your cap turns out to be, write the abstract last and give away the results.
The limits five institutions actually publish
APA is the number most students are handed, and its own Abstract and Keywords Guide gives it in five words: "Limit your abstract to 250 words." The same page opens with a caveat worth reading before you write anything, that "Abstracts and keywords are not typically required for student papers" and that students "should consult their instructor or institution to determine whether to include an abstract".
Past that, the cap belongs to whoever is collecting the document, and the spread is wider than the usual "150 to 300" suggests. Five sets of published rules, in their own words.
| Who sets it | The cap | In their words |
|---|---|---|
| APA 7 | 250 words | "Limit your abstract to 250 words." |
| University of Louisville | 150 for a thesis, 350 for a dissertation | "The thesis abstract must not exceed 150 words and the dissertation abstract must not exceed 350 words." |
| Arizona State University | 350 words | "All students must limit their abstracts to 350 words or less." |
| UC San Diego | 350 doctoral, 250 master's | "Abstracts of doctoral dissertations must not exceed 350 words; master's theses abstracts must not exceed 250 words." |
| Aberystwyth University | 250 to 300 words | "An abstract is a self-contained summary of 250-300 words of your thesis." |
Read down the middle column and the useful fact is the disagreement. One university caps a master's abstract at 250 and a doctoral one at 350. Another caps a thesis at 150, which is less than half, for a document of comparable size. There is no general number underneath any of this, only a stack of local ones, and the only one that can cost you marks is yours.
Where nobody hands you a number, the UNC Chapel Hill Writing Center's abstracts handout gives a proportion instead of a count: an informative abstract rarely runs more than about 10 percent of the whole paper. On a 4,000 word course paper that works out at 400 words, which is more room than any published cap above allows, so read it as a ceiling rather than a plan.
Check one more thing before you count anything, because it changes the arithmetic. Some universities count the abstract inside the assignment's own word limit and some exclude it, and one counts it while advising you not to write one at all. We worked through that element by element in do quotes count in the word count.
Whatever your cap is, it is a ceiling and not a target. A tight 180 words carrying all five parts below beats a padded 250, and if you cannot fit the five parts inside the limit, you are explaining too much rather than summarising too little. Cut the setup and keep the findings.
What an abstract is, and why you write it last
An abstract is a short, standalone summary of your whole paper, usually one paragraph, that lets a reader understand what you did and what you found without reading the rest. It sits at the top of the paper, right after the title, but it is not an introduction. An introduction sets up the problem and leaves the answer for later. Nothing is held back for suspense, because its job is to let a reader decide, in thirty seconds, whether the full paper is worth their time.
That self-contained quality is the rule that governs everything else. The abstract has to make sense on its own, pulled out of the paper and dropped into a search results page, which is exactly what databases do with it. So it carries no citations, no figure references, no "as discussed below," and no abbreviation a first-time reader would not recognize. If a sentence only makes sense to someone who has already read the paper, it does not belong in the abstract.
And you write it last. This is the single most common mistake: drafting the abstract early, when it feels like an outline of intentions. But an abstract summarizes a finished paper, and your aim sharpens, your method shifts, and your results are not final until the analysis is done. An abstract written first describes a paper that no longer exists by the time you submit. Write the body, the introduction, and the conclusion, and only then turn around and summarize what you actually wrote. We put the abstract at the end of the build for the same reason in how to write a research paper.
The five parts of an abstract
Whatever your field, a good abstract answers five questions in order. Think of it as one sentence or two per part, which is how 200 words holds an entire paper.
Background and problem
Open with the context and the gap. One or two sentences on why the topic matters and what is missing, unresolved, or contested in what is already known. This is not a literature review; it is the single problem your paper addresses, framed so a reader outside your exact subfield understands why anyone should care.
Aim
State what your paper set out to do. This is your research question or objective in one clean sentence: what you investigated, tested, or argued. If it still feels broad, how to write a research paper covers narrowing one. The aim follows directly from the gap you just named, so the reader sees the problem and your response to it back to back.
Method
Say how you did it. For an empirical paper, that is your design, participants or data, and approach, briefly: enough for a reader to judge how the results were produced, not a full methods section. For a non-empirical paper, it is your approach, the framework, the texts, or the kind of analysis you ran. Keep it to the essentials that make your results credible.
Key results
Report what you found. This is the part students underwrite, hiding the actual findings behind "results are discussed." Give the real outcome: the main result, the direction of the effect, the headline number if there is one. The results are why anyone reads the abstract, so they get the most weight. Report what the paper found, not what you hoped it would.
Conclusion
End with what it means. One sentence on the takeaway: the answer to your aim, why it matters, or what it implies for the field or for practice. This is where you state the significance without overclaiming. A conclusion that says more than your results support is the fastest way to lose a reader's trust in the first paragraph.
If you are staring at a blank box, fill in this one-sentence-per-part skeleton and then smooth the joins. It is the fastest way from a finished paper to a first abstract draft.
Background. [Topic] matters because [why], yet [what is still unknown or contested].
Aim. This [paper/study] [asks/tests/examines] whether [your question].
Method. We [design] with [participants or data], [measuring or analyzing] [what].
Results. We found that [main finding], with [the direction or headline number].
Conclusion. This suggests [what it means], which matters for [whom or what].
Structured versus unstructured abstracts
The five parts stay the same; what changes is whether they wear labels. An unstructured abstract is one flowing paragraph with no headings, and it is the default for most humanities, social science, and undergraduate work. The five parts are still there, just stitched into prose. This guide's worked example below is unstructured.
A structured abstract breaks the same content into labeled sections, often Background, Methods, Results, and Conclusions, each as its own short block. Plenty of empirical journals require it, across the sciences, engineering, and management, and the labels themselves vary by field: some ask for Purpose, Design, Findings, and Originality instead. Assignment briefs sometimes ask for it too. If yours does, follow the exact headings it specifies, because the labels are not optional decoration; they are the format. The content you write is identical either way. The only question is whether your guidelines want the seams showing.
Descriptive versus informative abstracts
There is a second distinction worth knowing, because the wrong one fails the assignment. A descriptive abstract says what the paper is about without giving away the findings: it names the topic, the scope, and the purpose, but stops short of the results and conclusion. It is short, often under 100 words, and you will mostly meet it in the humanities or as a conference proposal.
An informative abstract gives the whole picture, results and conclusion included, and it is what almost every research paper and journal expects. When an assignment says "write an abstract" with no other detail, it nearly always means the informative kind. If you write a descriptive abstract by accident, leaving out your findings, it reads as if you never reached any, which is the opposite of what you want. The five-part structure above is informative by design.
A worked example
Here is an informative, unstructured abstract for an invented student study, built from the five parts. The topic and findings are illustrative, not drawn from any real paper, so use it as a shape to copy, not a source to cite.
Students increasingly study with background music, yet whether it helps or harms learning remains contested, and little of the existing work separates familiar from unfamiliar tracks. This study examined whether the familiarity of background music affects reading comprehension in undergraduates. Sixty participants read matched passages under three conditions, silence, familiar music, and unfamiliar music, and completed a comprehension test after each. Comprehension scores were highest in silence and lowest with familiar music, while unfamiliar music fell in between, a pattern consistent with familiar tracks drawing more attention away from the text. The findings suggest that for demanding reading, silence outperforms music, and that when students do listen, unfamiliar music is the less costly choice.
Read it back against the five parts. The first sentence is background and the gap (familiarity is understudied). The second is the aim. The third is the method, design and sample in one line. The fourth is the key results, with the actual direction of the effect rather than "results are reported." The last is the conclusion, stating what it means without claiming more than the study showed. No citations, no abbreviations, no reference to figures, and it stands completely on its own. That is the whole job.
Keywords
Many journals and some assignments ask for a short list of keywords, usually three to six, placed just under the abstract. They are not decoration; they are how your paper gets found, the terms a database indexes and a searcher types. Choose the words a reader looking for your work would actually use: the core concepts, the method, the population, the field. Skip words already in your title where you can, since those are indexed anyway, and prefer the standard term over a clever one. For the example above, "background music, reading comprehension, attention, undergraduates, familiarity" would do the job.
Common mistakes
Most weak abstracts share a short list of fixable problems:
- Including citations or reference markers. An abstract is standalone. Leave the citations, footnotes, and "[3]" markers in the body where they belong.
- Being vague about results. "Findings are discussed" tells the reader nothing. State the actual result, including the number or direction if there is one.
- Writing it first. An abstract drafted before the paper describes a paper you did not write. Save it for last.
- Treating it as an introduction. An intro withholds the answer; an abstract gives it away. Include your results and conclusion.
- Going over the word limit. The cap is a rule, not a suggestion. Cut the setup before you cut the findings.
- Unexplained abbreviations or jargon. The abstract reaches readers outside your subfield. Spell things out or leave them for the body.
- Claiming more than the paper shows. An overreaching conclusion undermines the work in its most-read sentence. Match the claim to the evidence.
Draft the abstract from the paper you wrote
The abstract is the one part of the paper that is genuinely easier to automate, because it is a faithful summary of work you already did rather than an argument you have to think up. CiteOwl drafts the abstract from your finished paper, last, the way you should: it reads what you actually wrote and produces a summary that hits the five parts, and that draft arrives as a reviewable change you accept, edit, or reject like any other. It does not invent a finding that is missing from your results or smuggle in a citation, because the abstract is built from the paper in front of it, not from memory.
That is the same verify-first principle that governs the rest of the tool. CiteOwl searches actual literature and reads what it finds, so the sources behind your paper are real, and every change it makes, the abstract included, is a diff you review with version history to step back. You keep the argument and the final words; the tool handles the slow, mechanical parts and shows its work. If your next job is the survey of the field that feeds the paper, here is how to write a literature review with AI.