How to Read a Research Paper | A Practical 3-Pass Guide

How to read a research paper

It is a skill you can learn without understanding every sentence on the first try. Instead of reading a paper from beginning to end at the same speed, use a few passes: first decide whether it is relevant, then understand its argument, and only then examine the details you truly need.

This guide adapts the three-pass approach described by computer scientist Srinivasan Keshav and combines it with practical source-evaluation questions. It is designed for school, college and university students who need to read studies efficiently, prepare assignments, or begin a literature review.

Before you start: know what you need from the paper

Begin with a question. Are you looking for background information, evidence for an assignment, a method you can reuse, or a study to compare with other research? Your purpose determines how deeply you need to read. A paper that is only loosely related to your topic may need a quick scan, while a study that supports your main argument deserves closer attention.

Also identify the type of article. A primary research paper reports a study or experiment carried out by its authors. A review paper summarizes and compares earlier studies. Both can be useful, but they answer different questions: primary research gives you the original methods and results, while a good review can help you understand the wider field.

The three-pass method for reading a research paper

Keshav’s method recommends up to three passes rather than one slow, linear read. You can stop after any pass if the paper does not meet your needs. The time estimates below are starting points, not rules; a complex paper or a new subject may take longer.

Pass 1: Scan for relevance (about 5–10 minutes)

Do not start by struggling through every paragraph. Read the title, abstract and introduction, then scan the section headings, figures and conclusion. Glance at the references to see whether the authors cite work you recognize or whether there are useful sources to follow up.

At the end of this pass, try to answer five questions:

• Category: What kind of paper is this—an experiment, survey, analysis, review or description of a system?
• Context: What problem does it address, and how does it relate to your topic?
• Claim: What is the main question or conclusion?
• Contribution: What does the paper add beyond what was already known?
• Clarity: Can you follow the structure well enough to justify reading further?

If the topic is not relevant, the abstract does not match your question, or the paper is too preliminary for your assignment, record why and move on. That is a useful decision, not wasted effort.

Pass 2: Understand the argument and evidence

If the paper looks useful, read it more carefully. Focus on the research question, the method, the main results and how the authors interpret those results. Look closely at figures and tables: read their titles, labels, units, sample sizes and captions rather than treating a graph as decoration.

For an experiment or survey, ask who or what was studied, how participants or data were selected, what was measured, and what comparison was made. For a theoretical or technical paper, identify the assumptions and the steps that connect the argument to its conclusion. You do not need to solve every equation immediately; first understand what role it plays.

After reading, summarize the paper in two or three sentences without copying the abstract. If you cannot explain the main claim and the evidence in your own words, return to the relevant section or mark the point that remains unclear.

Pass 3: Examine the details you must rely on

Use the deepest pass for a paper you will cite as central evidence, reproduce, critique or build on. Check the methods and supplementary material, inspect the statistical or technical details relevant to the claim, and compare the authors’ conclusions with what the results actually show. Note assumptions, missing comparisons, small or unusual samples, and limitations the authors acknowledge.

Try to separate three things in your notes: what the researchers observed, how they interpret it, and what you think the evidence supports. A study can report a real association without proving that one factor caused another. The design of the study determines which conclusions are justified.

How to evaluate whether a research paper is trustworthy

A polished article or a journal name alone cannot guarantee that every conclusion is correct. Use a repeatable checklist, and read the paper in the context of other credible work. University guidance from Purdue OWL’s source-evaluation resources also emphasizes checking authority, purpose, evidence and the reliability of information.

• Question and method: Does the method actually answer the research question?
• Sample and data: Is the sample suitable for the conclusion? Are important exclusions or missing data explained?
• Evidence: Do the tables, figures and reported results support the main claim?
• Limitations: Does the paper discuss uncertainty, weaknesses or alternative explanations?
• Transparency: Can you find enough information about methods, data or materials to understand how the work was done?
• Context: Do other credible studies agree, disagree or add important qualifications?
• Interests and funding: Are funding sources or potential conflicts disclosed where relevant?

Peer review can improve a paper, but it is not a guarantee that the findings are error-free or that the study applies to every population. Check whether a paper has a correction, expression of concern or retraction when you are relying on an important result. For unfamiliar subjects, a university library or librarian can help you find suitable databases and distinguish scholarly sources from popular summaries; see Cornell University Library’s guide to finding articles.

A simple note-taking template for every paper

Keep your notes short enough to review. A consistent template also makes it easier to compare studies when you begin an assignment or literature review.

• Full citation: author names, year, title, journal or publisher, and DOI or stable URL.
• My question: why I opened this paper and what I need to learn from it.
• Main claim: the key point in my own words.
• Method and sample: what was studied and how.
• Best evidence: the result, table or figure that most directly supports the claim.
• Limitations: what the study cannot show or what remains uncertain.
• Use in my work: how it supports, challenges or adds context to my question.
• Follow-up: terms to look up and references worth reading next.

Record page numbers for direct quotations and use quotation marks in your notes. When you paraphrase, close the source and explain the idea in your own wording, then compare your version with the original to ensure you have not copied its sentence structure. Cite the source in your assignment even when you paraphrase.

Common research-paper reading mistakes to avoid

• Reading every paper equally deeply: use the first pass to decide where your time matters most.
• Trusting the abstract alone: verify the method, results and limitations before relying on a claim.
• Confusing correlation with causation: check whether the study design supports a causal conclusion.
• Highlighting without thinking: write questions and short summaries instead of marking entire pages.
• Ignoring contradictory evidence: look for studies that reach different results and consider why.
• Copying notes too closely: keep source details and page numbers so you can cite accurately later.

Frequently asked questions

How long should it take to read a research paper?

A first scan may take about 5–10 minutes. A careful read can take much longer, especially for technical methods or unfamiliar topics. Spend time in proportion to how important the paper is to your work.

Should I read the abstract or introduction first?

Start with the title and abstract to check relevance, then read the introduction for the research question and context. Scan the headings and conclusion before deciding whether to read the full paper closely.

What if I do not understand the methods section?

Write down the unfamiliar terms, look up the basic method in a trusted textbook or university resource, and return to the paper. If the method is central to a claim you plan to cite, do not treat it as understood until you can explain its purpose and limits.

Are review papers better than original studies?

Neither is automatically better. Reviews help map a topic and compare findings; original studies provide the methods and results of individual investigations. Use the type that fits your question, and consult original studies when a specific result matters.

Can AI help me read research papers?

AI tools can help explain terminology or organize questions, but they may misread tables, omit limitations or invent details. Check every important claim against the paper itself and cite the original research rather than an AI summary. Paper Bloom also has a separate guide to AI tools for research papers; use tools as support, not as a substitute for reading evidence.

Final thoughts

The most useful way to read a research paper is to match your reading depth to your purpose. Scan first, examine the argument and evidence second, and investigate the details only when the paper is important enough to justify the effort. Keep notes in your own words, check limitations, and compare claims with credible sources. These habits make academic reading more efficient without sacrificing careful judgment.

Sources and further reading

• Srinivasan Keshav, “How to Read a Paper,” ACM SIGCOMM Computer Communication Review (2007).
• Purdue Online Writing Lab: Evaluating Sources.
• Cornell University Library: Finding Articles.

Read for more information

Effective Ways to Study | 9 Evidence-Informed Methods That Help