How to review a research paper

A step-by-step method

Reviewing a paper well is a skill you can learn, and it improves your own writing. This method works for journal reviews, conference reviews and reviewing a colleague’s draft.

Try the AI peer review free

6 min read

Read it three times

  1. First pass: the big picture

    Title, abstract, introduction, figures and conclusion. What is the question, and what is the claimed answer?

  2. Second pass: the evidence

    Methods and results in detail. Does the design answer the question? Do the results support each claim?

  3. Third pass: the details

    References, figures, statistics, clarity and anything missing.

What to check

  • Contribution: is it new, and is the difference from the closest work clear?
  • Methods: appropriate, described well enough to repeat, with suitable controls or comparisons.
  • Results: do the data support the conclusions, and are uncertainties reported?
  • Literature: are key and recent works cited and discussed fairly?
  • Presentation: is the paper clear, and do the figures help?

Structure the report

  1. Summary

    Two or three sentences in your own words: shows the authors you understood the paper.

  2. Major issues

    Problems that affect the conclusions, each with why it matters and what would fix it.

  3. Minor issues

    Clarity, small errors, missing details.

  4. Recommendation

    For the editor, with a short justification, if the journal asks for one.

Tone

Review the paper, not the authors. Be specific, suggest fixes, and acknowledge what works. A good review is one the authors can act on, even if the decision is to reject.

Questions

Can students review papers?

Often yes, with a supervisor (co-review) or for student journals and conferences. Ask your supervisor; many journals welcome early-career reviewers.

Can I use AI to review a paper I was sent?

Check the journal’s policy first. Many forbid uploading confidential manuscripts to AI tools.