I've been asked to review a paper
Write a fair, useful review for a journal, even if it's your first and you feel underqualified.
In short
Start here, today. Open the invitation and the journal's reviewer guidelines. Write down the deadline, the report format and any conflict of interest you might have, then decide whether to accept. This takes about 20 minutes.
After that, work through these steps. Choose one to jump to it.
Step by step
Tick each step as you finish it. Open a step for the how.
0 of 6 steps done
Decide whether to accept
Say yes if you can assess at least the main parts of the paper, have no conflict of interest and have time to finish by the deadline. Journals often ask for reports within a few weeks, and Nature, for example, normally asks for two. If you can't do it, decline quickly and suggest someone else if you can. Editors would much rather have a fast no than a late yes.
Read once for the big picture
On the first read, don't note details. Ask what the paper claims, whether the question matters and whether the design and data could support the conclusions. Then write two or three sentences summarizing the paper in your own words. If you can't, that's useful feedback in itself.
Read again for the evidence
On the second read, look closely at the methods, analysis, figures and any supplementary material. Check whether the statistics suit the data, whether other explanations are considered and whether earlier work is cited fairly. Note anything you couldn't judge because it's outside your expertise.
Structure the report
Start with a short summary of the paper and its contribution. Then give your major concerns, the issues that affect whether the conclusions hold, followed by minor points such as unclear figures or missing references. Number your comments so the authors can respond to each one. Put anything sensitive in the confidential comments to the editor, and make sure those never contradict what you tell the authors.
Write the review you'd want to receive
Comment on the paper, not the people. Say what works as well as what doesn't, and for each concern explain why it matters and what might address it. A good test is whether you'd be comfortable if the authors knew you wrote it.
Respect confidentiality and the journal's rules
Don't share the manuscript or use its findings before it's published. Many journals ask reviewers not to upload manuscripts into generative AI tools and to declare any AI use in the report. If your supervisor involves you in a review, ask them to tell the editor, since journals often ask for the names of anyone who helped.
Words you can use
Is this you? Signs to look for
- An invitation to review has arrived and you're not sure you're senior enough to say yes.
- Your supervisor has handed you a manuscript and asked for your thoughts.
- You've read the paper twice and can't tell whether your concerns are serious or just picky.
- You're tempted to rewrite the study the way you would have done it.
Why this happens
Editors invite reviewers because they know something about the topic or the methods, not because of their job title. If the paper sits close to your own work, you may know the recent literature better than many senior colleagues. Feeling underqualified the first time is common, and the answer is to be clear in your report about which parts you could assess and which you couldn't.
A review has two readers. The editor needs enough information to make a decision, and the authors need to understand what would make the work stronger. Neither is helped by a list of personal preferences or a description of how you would have done the study. Specific comments tied to the paper's claims and evidence are the most useful thing you can give.
Reviewing also comes with obligations: keeping the manuscript confidential, declaring conflicts of interest and meeting the deadline. Journals set these out in their reviewer guidelines, and they're worth reading before you accept.
Worth a look (7)
- How to write a report, NatureA clear list of what a referee report should cover. Most of it applies to any journal.
- Policies and processes for referees, NatureConfidentiality, competing interests, timelines and consulting colleagues.
- Editorial policy on AI, NatureOne major publisher's rules on AI use, including for peer reviewers. Check your journal's own policy too.
- COPE, Committee on Publication EthicsGuidance on publication ethics, including for peer reviewers.
- Think. Check. Submit.Simple checklists to help you judge whether a journal or book publisher can be trusted before you submit. They're available in many languages.
- Open policy finderCheck a journal's open access policies, including which version of your paper you can share and where. It replaced Sherpa Romeo, so older guides may still use that name.
- DOAJ (Directory of Open Access Journals)A curated list of peer-reviewed open access journals in every field. Use it to find journals to read and to check that a journal you're considering is legitimate.
When to bring someone in
Your supervisor or another experienced researcher can read a draft of your report and tell you whether the tone and level of detail are right, as long as the editor agrees to that person seeing the manuscript. If you notice something that looks like plagiarism, duplicated images or fabricated data, raise it with the editor in the confidential comments rather than with the authors. Questions about the process can go straight to the handling editor.
Find the right person, or talk to Ify