How to Write a Research Paper Abstract

The 200 words that decide whether your paper gets read, the four-move structure, what to include, what to cut, and writing it last.

The abstract is the most-read and least-carefully-written part of most papers. Editors screen on it, databases index it, and readers decide from it whether to continue. In roughly 200 words it must convey the entire study, which is a discipline of compression, not a teaser.

The Four Moves

Purpose/problem: what question the study addresses and why it matters. One or two sentences.
Methods: how you investigated it, design, participants, key measures.
Results: the actual findings, with specific numbers for quantitative work. This is where most abstracts fail, "results are presented" tells the reader nothing.
Conclusion/implication: what the findings mean and why they matter.

Structured vs Unstructured

Many journals, especially in medicine and the sciences, require a structured abstract with explicit headings (Background, Methods, Results, Conclusions). Humanities and social-science journals often use a single unstructured paragraph. Follow your target journal's format exactly. It is a submission requirement, not a preference.

Be Specific, Especially in Results

Vague abstracts read as if the authors are hiding a thin result. Compare "the intervention improved outcomes" with "the intervention reduced readmissions by 18% (p = .002)". The second earns the reader's trust and tells them whether the paper is relevant.

What to Leave Out

Citations (the abstract stands alone).
Abbreviations that need defining.
Information not in the main text.
General background beyond the one framing sentence.

Keywords

Choose keywords that maximise discoverability, the terms a searcher would actually type, including recognised subject headings (e.g. MeSH) where the field uses them. Poor keywords bury a good paper.

Write It Last

An abstract summarises what the paper actually says, not what you intended it to say. Draft it after the paper is complete, then check every claim in the abstract appears in the body.

Checklist

Does it cover purpose, methods, results and implication?
Are results specific, with numbers where relevant?
Does it match the journal's required format?
Is every abstract claim supported in the main text?
Is it within the word limit, counted rather than estimated?
Would the abstract still make sense to someone who reads nothing else?

Word Budget for the Four Moves

Most abstracts fail by spending half the budget on background. A rough allocation keeps the proportions honest, and you can adjust from there.

Move	Share of a 250-word abstract	Typical sentences
Purpose and problem	15 to 20 per cent	1 to 2
Methods	20 to 25 per cent	2 to 3
Results	35 to 40 per cent	3 to 4
Conclusion and implication	15 to 20 per cent	1 to 2

If your results section is not the largest block, the abstract is probably promising rather than reporting.

Before and After

The difference between a weak and a strong abstract is almost always specificity. Compare these two versions of the same study.

Weak. This study investigates the important issue of student engagement in online learning, which has attracted considerable attention in recent years. A range of data was collected from university students and analysed using appropriate methods. The results are presented and discussed, and a number of interesting findings emerged. Implications for practice are considered.

Strong. Online delivery expanded rapidly after 2020, but its effect on sustained engagement is unclear. This study examined whether synchronous seminars predict module completion among 412 undergraduates at one UK university across two semesters. Logistic regression showed that attending more than half of synchronous sessions was associated with higher completion (OR 2.4, 95% CI 1.6 to 3.6). Interview data from 18 students indicated that timetabled sessions functioned as commitment devices rather than as preferred learning formats. Institutions may gain more from scheduling structure than from adding asynchronous content.

The weak version could describe thousands of studies. The strong one could describe only this one. That is the test.

Verb Tense, Move by Move

Move	Tense	Example
Background	Present	"Completion rates vary widely across delivery modes."
Purpose	Past or present	"This study examined whether..."
Methods	Past	"Data were collected from 412 undergraduates."
Results	Past	"Attendance predicted completion."
Conclusion	Present	"Scheduling structure appears to matter more than content volume."

Abstract Types

Type	Where used	Key feature
Informative	Most journals and dissertations	Reports the actual findings
Descriptive	Some humanities journals	Says what the paper covers without giving results
Structured	Medicine, health, many sciences	Explicit headings, often with word limits per section
Graphical	Some science journals, alongside text	A single figure summarising the study
Conference	Submission for talks and posters	Often shorter, and may be reviewed before results exist

Dissertation abstracts differ from journal abstracts in one respect worth noting. They usually summarise a much longer piece, so the temptation to cover every chapter is strong. Resist it. The abstract still reports the study, not the table of contents.

Common Mistakes and Fixes

Mistake	Fix
"Results will be discussed"	State the results. This phrase belongs only in a pre-results conference abstract
Three sentences of background	Cut to one. The reader wants your study, not the field
Citations in the abstract	Remove them. The abstract must stand alone
Undefined abbreviations	Spell out, or drop the abbreviation entirely
A claim that appears nowhere in the paper	Check every sentence against the body. This is a common source of reviewer irritation
Over the word limit	Count it. Editors reject on this without reading further

Compressing Without Losing the Study

Getting from a 400-word draft to a 250-word abstract feels like choosing what to lose. It is usually not. Most of the excess is structural padding rather than content, and cutting in a fixed order removes it without touching the substance.

Start with throat-clearing. Phrases such as "it is widely recognised that", "this paper seeks to", "in order to" and "a number of" carry no information. Removing them typically recovers thirty words on its own and makes the prose read faster.

Next, cut the second background sentence. Almost every abstract has one sentence establishing the field and a second elaborating it. The second is nearly always dispensable, because the reader who needs it will read the introduction.

Then compress the method into its identifying features. A reader needs the design, the sample and the analytic approach. They do not need your inclusion criteria, your software, or the number of items in each subscale, all of which live in the methods chapter.

Only after those three passes should you touch the results, and even then, prefer combining findings into one sentence over deleting one. A result that disappears from the abstract effectively disappears from the literature, because most readers will never see the full text.

Finally, count the words rather than trusting the word processor's estimate of what counts. Some journals include the structured headings in the limit and some do not, which is exactly the sort of detail that gets a submission returned before review.

Keywords That Actually Get You Found

Keywords are an indexing decision, not a summary. Choose the terms a reader would type, avoid repeating words already in your title where the database indexes both, and use recognised subject headings where your field has them, such as MeSH in the health sciences. Three to six specific terms beat eight vague ones.

Where Ethical Support Fits

Asking a supervisor whether your abstract represents the study accurately is standard practice, and so is a proofread for concision. Both are about expression, not authorship.

What stays yours is the study and the claims. An abstract that overstates the findings is a problem no editing can fix, so make sure the numbers in it match the numbers in your results chapter exactly.

Frequently Asked Questions

How long should an abstract be?

Follow the limit you were given. Journals commonly set 150 to 300 words. Dissertations often allow around 300. Where no limit is stated, 250 is a safe target.

Can I include a table or figure?

Only where a graphical abstract is explicitly invited. Otherwise the abstract is continuous text.

Should I write it first or last?

Last. Some people find drafting a rough version early useful for keeping the paper focused, but the submitted version must describe what the finished paper actually says.

Does the abstract count toward my word limit?

For dissertations, usually not, but this varies by department. Check the handbook rather than assuming.

Can I reuse sentences from my introduction?

Better not to. Repeating your own phrasing verbatim reads flatly, and text-matching software will flag the overlap even though it is internal.

What about an abstract for a qualitative study?

The same four moves apply. Instead of effect sizes, report what the analysis found, naming themes or the central argument. Vagueness is a bigger risk here, so name the findings rather than gesturing at them.

Your Next Step Today

Highlight the results sentences in your current abstract. If they occupy less than a third of the text, or contain no specific finding, rewrite them first. That one change does more for an abstract than any amount of polishing elsewhere.

Trusted Sources

University of Southern California, The Abstract, in Organizing Your Social Sciences Research Paper. Accessed 11 August 2026.
University of Wisconsin-Madison Writing Center, Writing an Abstract for Your Research Paper. Accessed 11 August 2026.
University of Southern California, The Results. Accessed 11 August 2026.
EQUATOR Network, for reporting guidelines that specify structured abstract content by study type. Accessed 11 August 2026.

Abstract requirements are set by your journal or department. Where this guide and their instructions differ, follow theirs.