TL;DR: A research paper outline is a plan for your paper: it lists each section in order, with the main point and evidence you'll use in each one. For a class paper built on sources, that means an introduction with a thesis, a few body sections, and a conclusion. If you collected your own data, your outline should follow IMRaD: Introduction, Methods, Results, and Discussion. And if your teacher, science fair, or journal gives you a format, just use theirs.
If you're a high schooler with a research paper coming up, an outline is the best place to start. A good outline means you'll know what goes in every paragraph before you write the first one. You can use an outline for history papers, interview studies, and projects from our list of STEM research ideas.
What is a research paper outline?
A research paper outline is a map of your whole paper that you write before your first draft. Your outline should answer three questions.
Main idea: What's your question, thesis, or hypothesis? A thesis is a position you argue, and a hypothesis is a result you predict.
Order: What goes in each section?
Evidence: Which source or data backs up each point?
Many students start from one of two templates: the class paper, which argues from sources, and IMRaD, which reports your own data. IMRaD stands for Introduction, Methods, Results, and Discussion. If you interviewed people or built something in code, you'll adjust one of these, as the section after the templates shows.
Research paper outline template for a class paper
A class paper outline follows the alphanumeric format (I, A, 1, a), with Roman numerals for the main sections. You can use this template whenever you argue a position from sources, as in most history papers. Purdue OWL, Purdue University's online writing lab, describes it as one of three outline formats.
I. Introduction
A. Hook: a specific fact or case from one of your sources
B. Background: what your reader needs to know
C. Thesis: your position in one sentence
II. Body sections, one for each main point
A. Claim: the section's topic sentence
B. Evidence: quotes or paraphrases from your sources, with citations
C. Link: how this evidence supports your thesis
III. Counterargument
A. Objection: the strongest argument against your thesis, stated fairly
B. Response: your answer, with evidence
IV. Conclusion
A. Thesis, restated: what your evidence has shown
B. What's next: what's still unknown
V. Works cited: in your teacher's citation style
Research paper outline template for an experiment or data analysis (IMRaD)
An IMRaD outline uses the decimal format (1, 1.1, 1.2), so you can see which point belongs under which section. You can use this template to report data you collected or analyzed yourself.
Title: what you tested, and on what
Abstract: a short summary of your purpose, method, key result, and conclusion (write it last, and only if your rules ask for one)
1 Introduction
1.1 Problem: why it matters, with one or two sources
1.2 Earlier work: what other researchers found
1.3 Gap: what nobody knows yet
1.4 Question and hypothesis: one sentence each
2 Methods
2.1 Data or materials: what you used, and where it came from
2.2 Procedure or model: each step, clear enough to repeat
2.3 Comparison and measurement: what you compared, and how you scored it
3 Results
3.1 Main result: the answer to your question, with Table 1
3.2 Other results: your figures, plus any surprises
4 Discussion
4.1 Answer: whether your hypothesis held up, compared with earlier work
4.2 Limitations and next steps: what you couldn't test, and what you'd try next
5 References and acknowledgments
How to structure a research paper outline for your type of research
You'll pick your outline's structure based on the kind of research you did. The first two kinds use the templates above, and the other two adjust IMRaD.
Argument from sources: use the class paper template above. You state a thesis, back each claim with evidence, answer a counterargument, and conclude.
Experiment or data analysis: use the IMRaD template above.
Interviews or observation (qualitative research): keep IMRaD, explain in Methods how you chose people and found themes, and give each theme a short quote in Results. The American Psychological Association's qualitative research standards list what each section should report.
Coding or engineering project: adjust IMRaD. Your Introduction states the problem, your Methods describe your design and how you built and tested it, and your Results show how well it worked, as in the University of Melbourne's guide to design reports.
How to write a research paper outline in 7 steps
To build a strong outline, start from your assignment's rules and a one-sentence answer to your question. Here are 7 steps to build one.
Check what your paper needs. If your teacher, science fair, or journal requires a format, follow it before any template. Write down the sections, length, and citation style they ask for.
Write your question and a one-sentence answer. That answer is your thesis or hypothesis. If it won't fit in one sentence, narrow your topic.
Copy the template or structure that fits your research.
Add at least two points under each heading, as Purdue OWL advises.
Attach a source or data to every point. List only sources you've read. With Semantic Scholar, a search tool for research papers, you can find papers and see who cites them.
For original research, plan your Results. List each table, figure, or theme.
Test your outline. Read just the top-level points. If they don't answer your question in order, rearrange them. For example, say your Introduction asks whether augmentation helps more with less data. Your Discussion should answer that same question.
It's much faster to fix an outline than a whole draft, so show yours to a teacher or mentor.
Research paper outline example for an AI project
Here's the IMRaD template filled in for an AI project you could run with basic Python in Google Colab, an online coding notebook. In a strong AI project outline, every line is specific enough to start working on today.
Title: Does Data Augmentation Help an Image Classifier More With Less Training Data?
1 Introduction
1.1 Problem: Image classifiers are AI models that sort pictures into groups. They need lots of labeled images, which take time to make. [Cite sources.]
1.2 Earlier work: Data augmentation means making extra training images by flipping and cropping the ones you have. [Cite papers.]
1.3 Question and hypothesis: Does augmentation help a small CNN (an AI model built for images) more when it trains on 10% of the data? Hypothesis: yes.
2 Methods
2.1 Data: CIFAR-10, 60,000 small color images in 10 classes, with 50,000 for training and 10,000 for testing.
2.2 Comparison: four conditions, using 10% or 100% of the training images, with augmentation on or off. Run each one three times and report the average test accuracy.
3 Results: Table 1, with test accuracy for all four conditions [fill in after running]
4 Discussion: whether the gain was bigger at 10%, plus limits (one dataset, one model) and next steps
5 References: [the CIFAR-10 technical report and every paper you cited]
To train a model like this yourself, you can learn Python from the start in Veritas AI's AI Scholars program. At the end of Veritas AI's AI Scholars program, you build a group AI project with three or four classmates.
How to outline a paper for a science fair or student journal
Science fairs and student journals each set their own rules, so you can start from IMRaD and adjust it to theirs. Here are two examples, as of October 2026.
Regeneron ISEF, the International Science and Engineering Fair: your abstract can be up to 250 words. Split your outline's abstract slot into your purpose, procedure, key results, and conclusions. Leave out tables, images, and a bibliography.
Journal of Emerging Investigators (JEI), a science journal for students: JEI lists Materials and Methods after the Discussion, so move Methods to the end of your outline. Keep your summary under 250 words and your title to 110 characters or fewer.
Common research paper outline mistakes
Three common outline mistakes are writing topics instead of points, leaving points without evidence, and mixing your results with what they mean. Here's how to fix each one, using the AI example.
Writing topics instead of points. A topic only names a subject. A point is a claim you can test or support. Before: "Data augmentation." After: "Augmentation helps more when you have less data."
Leaving points without evidence. A point with nothing behind it usually turns into a vague paragraph, so note the source or data you'll use for each one. Before: "Labeled images take time to make." After: "Labeled images take time to make [cite the source you read]."
Mixing your results with what they mean. In your Results section, you say what you found. In your Discussion, you say why it matters. Before, under Results: "Augmentation raised accuracy, so small datasets need it." After, under Results: "Augmentation raised accuracy more with 10% of the data (Table 1)." Under Discussion: "Augmentation may matter most when data is scarce."
FAQs
How long should a research paper outline be?
There's no set length. Your outline is long enough when every point in your paper appears in it with its evidence.
What are the 7 steps of writing a research paper?
You choose a topic, read your sources, and write your thesis or hypothesis. Then you outline, draft, revise, and format your citations. When you draft, the San Diego State University Writing Center suggests writing your introduction last. You outline in the middle of the research paper process, turning your reading into a plan for your draft. For an AI project, you can also follow these tips on writing an AI research paper in high school.
Can high school students publish a research paper?
Yes. Many high schoolers publish their research in student journals, which review papers written by students. You can compare your options, from how selective each journal is to how to submit, in our guide to high school research journals. With some Python under your belt, you can design your own AI research project with a mentor, one-on-one for 15 sessions, in Veritas AI's AI Fellowship. In four extra sessions, a publication specialist helps you prepare a journal or competition submission.
Official pages to check
You can check each rule on its official page.
Turn Your Outline Into Real AI Research With Veritas AI
If you'd like to run a study of your own, Veritas AI can guide you through an AI research project. Veritas AI has taught more than 3,200 students from over 160 countries. Students at Veritas AI plan and run a real research project, guided by mentors who include researchers from Anthropic, Meta, and NVIDIA. Of the papers Veritas AI students have submitted, 79% were accepted, according to Veritas AI internal data. You can look at the programs side by side to find your starting point.
About the author: Tyler Moulton is the General Manager of the Lumiere Research Scholar Program and a former Harvard and Cambridge researcher in AI and astronomy.