How long should each section of a research paper be?

In 58 published papers we measured in August 2026, the median split of the body text was introduction 14 percent, methods 26 percent, results 21 percent, discussion 23 percent and conclusion 4 percent, with the remaining tenth spent on extra sections such as related work. On a 3,000-word paper that is roughly 420 words of introduction, 780 of methods, 630 of results, 690 of discussion and 120 of conclusion. Treat those as starting allocations rather than rules: the papers varied widely, and a journal with a hard word limit compresses methods first. The number worth defending is the introduction, which was under a fifth of the body in 50 of the 58 papers.

What should each line of an outline contain?

A claim, the evidence for it, the source that carries the evidence, and a word budget. The claim is the point the section argues, written as a sentence rather than a topic. The evidence is the specific finding, number or passage you will use. The source is a real paper you have opened, not a placeholder. The budget is how many words the section gets, which is what stops one section eating the paper. An outline line with a claim and a source is a paragraph you can draft in one sitting. A line that names only a topic is a decision you have postponed.

What do I do if my paper does not fit the IMRaD shape?

Published papers bend it constantly. Of the 80 papers we measured, 11 ran results and discussion under a single heading, 17 had no separate results heading at all, and 18 carried a literature review or related work section that IMRaD does not name. If your findings only mean something once interpreted, combine results and discussion. If your work is a review, a design, a model or an argument rather than an experiment, replace methods and results with the shape your work actually has: what you built, how you tested it, what it does. Keep the underlying order, which is question, what you did, what you found, what it means.

What is the difference between alphanumeric and decimal outline format?

They are two ways of labelling the same hierarchy. Alphanumeric alternates Roman numerals, capital letters, Arabic numbers and lowercase letters as you go deeper: I, A, 1, a. Decimal nests numbers with dots: 1, 1.1, 1.1.1, so the label itself tells you the depth of every point. Decimal is standard in engineering and technical reports because numbered sections get cross-referenced in the text. Alphanumeric is the default in humanities and social science classes. Neither affects a single mark on the paper itself, so use whichever your assignment names and spend the saved effort on the word budget instead.

Research paper outline (with word counts per section)

The CiteOwl Team· Updated 11 August 2026· 15 min read

An outline is a word budget with headings on it. The headings are the easy part and every guide gives you the same ones. The part that decides whether the paper works is how many words each heading gets, because a paper that spends 40 percent of itself on the introduction has already failed no matter how good the sentences are. This page gives outlines for the four paper shapes students are actually set, a per-section word budget measured from 80 published papers rather than guessed at, and the decision procedure for when your work does not fit the standard shape, which is more often than the standard shape admits.

Contents

What published papers actually spend their words on
Your word budget, at three paper lengths
Outlines for the four shapes you get assigned
What each outline line has to carry
When your work does not fit the standard shape
Alphanumeric or decimal, and why it barely matters
The outline mistakes that cost the most
Filling the source lines without faking them

Most outline advice stops at the list of headings, which is the part you could have guessed. Nobody tells you that the introduction gets about a seventh of the paper, that the methods section is usually the longest thing in an empirical paper, or that roughly a fifth of published papers quietly ignore the structure everyone teaches. So we went and counted.

What published papers actually spend their words on

On 11 August 2026 we downloaded the full text of 80 research articles from PLOS ONE, indexed under the subject terms Climate change and Transportation and published since January 2024, and counted the words in every top-level section. Of those 80, 58 carried all four IMRaD headings separately, and those 58 are the basis of the split below. Anyone can rerun this: the articles are open access and the section headings are in the XML.

Section	Median share of the body	Middle half of papers	Median words
Introduction	13.9%	11.5% to 17.0%	842
Methods	25.9%	19.3% to 29.3%	1,308
Results	20.6%	16.2% to 29.1%	1,190
Discussion	23.2%	15.8% to 30.4%	1,232
Conclusion	3.6%	2.0% to 4.9%	190

The median body ran 5,912 words, with the middle half between 4,617 and 7,315. Four numbers in that table are worth more than the rest of this page.

The introduction is small. It was under a fifth of the body in 50 of the 58 papers. Student papers routinely run double that, because the introduction is the section you can write from general reading before you have any results, so it is where the words go while you are avoiding the harder sections.

Methods beat results. The methods section was longer than the results section in 33 of the 58 papers. That inverts what most students assume, and it is the single most useful thing on this page for anyone outlining an empirical report.

The conclusion is tiny. A median of 190 words, about 4 percent. If your outline has a conclusion the size of a body section, it is doing work that belongs in the discussion.

Nothing is tight. Look at the middle-half column: results ran anywhere from a sixth to nearly a third of the paper. These are starting allocations, not rules, and the honest way to use them is as a check on a first draft rather than a specification for it.

One caveat that matters, because it cuts in a known direction. PLOS ONE tells authors "manuscripts can be any length" and asks for methods with "enough detail to allow suitably skilled investigators to fully replicate your study" (PLOS ONE submission guidelines). A journal with a 3,000-word cap squeezes methods first, usually by pushing detail into supplementary material. So read the methods share here as the length methods reaches when nothing is stopping it, which is exactly the case in a thesis chapter or an unlimited-length assignment.

Your word budget, at three paper lengths

Take the medians, round them, and you have a budget you can put next to each heading before you draft a word. The shares do not add to 100 because about a tenth of a typical paper goes on sections IMRaD does not name, which the last row keeps honest.

Section	Share	1,500-word paper	3,000-word paper	6,000-word paper
Introduction	14%	210	420	840
Methods	26%	390	780	1,560
Results	21%	315	630	1,260
Discussion	23%	345	690	1,380
Conclusion	4%	60	120	240
Everything else	12%	180	360	720

Write the number beside the heading in your outline document and keep it there while you draft. It converts the vaguest question in writing, "is this section long enough?", into arithmetic. It also tells you something before you start: 315 words of results is three or four paragraphs, which is two findings well explained, not seven listed. If your outline has seven findings and a 1,500-word limit, the outline is wrong, and finding that out now costs a minute instead of a weekend.

An argued paper with no methods or results, the standard humanities or social science essay, redistributes that 47 percent across its body sections. What does not change is the introduction at roughly a seventh and the conclusion at well under a tenth. Those two are the ones students overspend, and they are overspent in every discipline.

Outlines for the four shapes you get assigned

Almost every undergraduate and taught-masters assignment is one of four shapes. Pick yours, take the skeleton, and put your budget beside each line.

The argued paper, built from sources

The default in humanities, law, politics and most social science coursework. The body is organised by sub-argument, never by source, and the order is a claim about which point earns the next.

Introduction (14%): the situation, the problem in it, your question, your thesis in one sentence. Body section per sub-argument (roughly equal shares of 60 to 65% between three and five of them): each opens with its claim, brings the evidence, then says what the evidence licenses. One body section carries the strongest objection to your thesis and answers it, which is the section markers look for and students skip. Conclusion (5%): the thesis, now earned, and the wider point. References.

The one structural decision is where the counter-argument goes. Put it last if your thesis is strong and the objection is a caveat. Put it second if the objection is what the reader arrives already believing, because everything after it is wasted otherwise.

The empirical report, IMRaD

Lab reports, survey projects, quantitative dissertations, most science coursework. The order is fixed by convention so any reader in the field can find each part without reading the paper.

Introduction (14%): what is known, the gap, the question or hypothesis. Methods (26%): design, participants or data, materials, procedure, analysis. Results (21%): what you found, with the figures and tables carrying most of it, and no interpretation. Discussion (23%): what it means against the question and the prior work, then limitations. Conclusion (4%). References.

The trap here is a results section written as prose describing tables. In the papers we measured the median article carried 5 figures and 3 tables, and in a well-built results section the prose points at them rather than restating them. Outline the figures first: one line per figure saying what it will show. If a figure has no sentence to earn, it is decoration; if a claim has no figure, it belongs in the discussion.

The literature review or review paper

Organised by theme, not by paper. Introduction (15%): the field, why the review is needed, the question the review answers. A methods or search section, short but real, saying which databases you searched, with what terms, over what dates, and what you included and excluded. Then three to five thematic sections (65%), each synthesising several sources into one point about the field. Then a gap section that names what the literature has not settled, which is where your own project gets its justification. Conclusion (5%).

The search section is the part students leave out and reviewers ask for. It is not bureaucracy: the PRISMA 2020 checklist, the standard systematic reviews are held to, spends four of its methods items on the search alone, items 5 to 8, covering eligibility criteria, information sources, the full search strategy, and how studies were selected (PRISMA 2020 checklist). You are not writing a systematic review, but two sentences in the same shape make the difference between a review a reader can trust and a reading list.

The engineering, design or computational report

The one IMRaD fits worst, and where students force it anyway. The real shape is: introduction and problem statement (15%), background and related work (15%), your approach, design or system (25%), evaluation, which is your methods and results welded together (30%), discussion and limitations (10%), conclusion (5%).

Two things separate a good one from a bad one. The related work section says what each prior approach could not do and your work can, so it sets the bar your evaluation has to clear. And the evaluation section names the baseline before it reports a number, because a result with no comparison is not a result. If your report compares nothing, the outline needs a baseline before it needs prose.

What each outline line has to carry

A heading with a topic under it is not an outline, it is a promise to decide later. Each line in a working outline carries four things.

The claim. A sentence, not a noun phrase. "Tree canopy cover reduces daytime street-level temperature more than reflective paving" is a claim. "Tree canopy" is a subject heading.
The evidence. The specific finding, figure or number you will use. Not "studies show" but the thing the study showed.
The source. A real paper you have opened, with the author, year and DOI written down. The moment to check a source exists is while it costs you nothing.
The budget. The word count from the table above. A section with no number attached is the one that eats the paper.

Here is one line done properly, from an outline for a paper on cooling city streets:

Body 2 (420 words). Claim: the cooling a street tree delivers depends on canopy geometry, not just canopy area, so "plant more trees" underspecifies the policy.
Evidence: the energy-balance account of why urban surfaces store and re-release heat at night, and what shading changes about it.
Source: Oke, T. R. (1982). The energetic basis of the urban heat island. Quarterly Journal of the Royal Meteorological Society, 108(455), 1–24. doi.org/10.1002/qj.49710845502
Gap: needs a recent empirical source for the geometry claim specifically, not the 1982 mechanism paper.

Notice the last line. Writing down what the section still needs is the difference between an outline you can hand to yourself on Thursday and one you have to re-derive. The gap note is also the honest version of a placeholder: it names what is missing instead of hiding it behind a citation you have not read.

When your work does not fit the standard shape

This is the section the template guides never write, and it is the most common reason an outline stalls. Published papers bend the standard shape constantly. In our 80 papers: 11 ran results and discussion under one heading, 17 had no separate results heading at all, 18 carried a literature review or related work section that IMRaD does not name, and 67 ended with a conclusion section that IMRaD does not name either. The structure everyone teaches is a median, not a rule.

So here is the decision procedure.

Your findings mean nothing until interpreted. Combine results and discussion into one section, ordered by finding. Common in qualitative work, where a quotation and its reading cannot be separated without saying everything twice. Check your brief first: some markers require them split.
You have no data of your own. Drop results. Methods becomes a short section describing your search or your corpus, and the body becomes thematic sections. This is a review, and the shape above fits it.
You built something rather than measured something. Replace methods and results with approach and evaluation. The reader still needs the same four things: what question, what you did, what came out, what it means.
You ran more than one study. Do not stack two IMRaD papers. Use a short general introduction, then Study 1 with its own compact methods, results and short discussion, then Study 2, then one general discussion that does the work of joining them. The general discussion is the paper; the studies are its evidence.
Your theory or framework needs explaining before the method makes sense. Give it its own section between introduction and methods, and keep it under 15 percent. If it runs longer, it is the paper and the empirical part is the illustration, which is a different outline.
Your brief specifies a structure. The brief wins, every time, over anything on this page. Where a department hands out a template, the marking rubric is usually built from it line by line.

Here is the opinion this page exists to argue. The alphanumeric-versus-decimal question that most outline guides devote half their length to is worth about thirty seconds of your attention, and the word budget is worth an hour, because no marker has ever deducted a point for the wrong kind of indent and every marker notices a paper that spends half its length clearing its throat. If you take one thing from this page, take the numbers in the second table and write them next to your headings.

Alphanumeric or decimal, and why it barely matters

Two labelling conventions organise the same hierarchy. Alphanumeric alternates Roman numerals, capital letters, Arabic numbers and lowercase letters as you go deeper: I, A, 1, a. Decimal nests numbers with dots, 1, 1.1, 1.1.1, so the label states the depth of every point. Purdue OWL's guide to outline types has a full sample of each if you want one to copy.

There is one real reason to prefer decimal, and it is not tidiness. In engineering and technical reports, sections get cross-referenced in the running text ("as shown in 3.2"), and a numbering scheme that survives that reference is worth having. Everywhere else the choice is cosmetic. Use what your assignment names, and if it names nothing, use whichever you will actually keep consistent for the length of the draft.

The related question, whether to write a topic outline (short phrases) or a sentence outline (full claims), is not cosmetic. A sentence outline forces you to know what each section argues before you write it, which is why it takes longer and why it is worth it for anything graded. A topic outline is fine for a short paper on ground you know.

The outline mistakes that cost the most

Organising the body by source. "What Lebreton found, then what Oke found" is a reading list. Group by claim and bring sources to it.
An introduction with no budget. The most expensive mistake on this page, and the measurement above is the argument: under a fifth of the body in 50 of 58 published papers.
Headings that name topics. A heading that names a subject gives the section no job. Name the point.
Placeholder sources. A citation you have not opened is a section that can collapse under checking. Write the gap note instead.
Drafting inside the outline. If a line runs to a paragraph, you are writing. Stop and go write.
Outlining once. An outline is a working document. When a section turns out to have no evidence behind it, change the outline rather than writing the section anyway.

If the thesis at the top of your outline still feels soft, sharpen it before you build sections under it with how to write a thesis statement, and if the question itself is still broad, how to write a research question covers the narrowing. For an empirical outline, the methods line has conventions of its own, in how to write a methodology.

Filling the source lines without faking them

The source line is where an outline is strongest and where it goes wrong quietest. An outline full of plausible references feels finished, and if those papers do not say what you attached to them, or do not exist, the structure is hollow. A general chatbot will fill those lines instantly with authors, journals and DOIs that look perfect and resolve to nothing, which is why AI makes up citations. An outline is exactly where an invented reference slips in unnoticed, dressed as settled background.

CiteOwl is built the other way round. It searches real literature, reads what it finds, and shows the quote behind every claim it attaches, so a source line in your outline is a paper you can open rather than a plausible string. Hand it a thesis and a section map and it can take the outline to a first cited draft, with every change arriving as a diff you accept or reject. The argument, the order and the budget stay yours. If you want to check the references already sitting in a draft, our free citation checker resolves them against CrossRef without an account.