How peer review works

Peer review is the quality check in academic publishing: independent experts in the field read a submitted manuscript, assess whether its methods are sound and its conclusions justified, and advise the editor on whether to accept, request revisions or reject it. The editor weighs these reports and makes the decision. Peer review does not guarantee that a paper is correct, but it catches many problems and usually makes the work clearer.

The stages of peer review

1. Submission and editorial check. The editorial office checks completeness and formatting. An editor then reads the manuscript and decides whether it fits the journal and merits review. Many papers are rejected at this stage, often because of fit.
2. Reviewer invitation. The editor invites typically two or more experts. Finding willing reviewers can take time.
3. Review. Reviewers read the manuscript, assess methods, results and interpretation, and write a report with comments for the authors and, often, confidential remarks for the editor.
4. Decision. Based on the reports, the editor decides: accept, minor revision, major revision or reject.
5. Revision and re-review. Authors revise and reply point by point. The revised manuscript may go back to the same reviewers.
6. Acceptance. Once the editor is satisfied, the paper moves into production.

The steps around submission are explained in how to publish a research paper.

Review models compared

Model	Who knows whom	Strengths	Weaknesses
Single-blind	Reviewers know the authors; authors do not know the reviewers	Widely used, simple	Reviewers may be influenced by authors’ names or institutions
Double-blind	Neither side knows the other	Reduces some forms of bias	Authors can sometimes be identified from the content
Open	Names of reviewers and authors are known to each other	Accountability, recognition for reviewers	Some reviewers may be less candid
Transparent	Review reports are published with the article	Readers see how the paper evolved	Not every reviewer wants reports published
Post-publication	Comments after publication, for example on preprints	Fast, open discussion	Uneven coverage and depth

Journals state their model in the author guidelines. Some also offer registered reports, where the study design is reviewed before data collection.

What reviewers look for

Reviewers usually ask:

• Is the research question clear and relevant to the journal’s readers?
• Is the method appropriate and described well enough to be repeated?
• Do the results support the conclusions, and are limitations acknowledged?
• Is the literature cited fairly and up to date?
• Are figures, tables and statistics correct and understandable?
• Are ethical requirements, data availability and conflicts of interest addressed?

Knowing these questions helps you anticipate criticism before submission.

How to respond to reviewer comments

• Read all reports, then wait a day before you draft a reply
• Create a response document that quotes each comment and gives your answer below it
• Say exactly what you changed and where (page, line or section)
• Thank reviewers once, briefly, without excessive praise
• Disagree only with evidence, politely and specifically
• Check that changes do not create new inconsistencies elsewhere in the paper

A “major revision” is not a rejection. It means the editor sees potential if the main concerns are addressed.

Peer review and AI tools

Reviewers receive unpublished work in confidence. For that reason, many publishers prohibit reviewers from uploading manuscripts or reports into generative AI tools, and some ask for disclosure of any AI assistance. Authors, in turn, are usually required to declare AI use in their manuscripts.

Note: Whether and how AI may be used, as an author or as a reviewer, is decided by the journal’s or publisher’s policy and, for students, by your examination regulations and supervisor. AI use often has to be disclosed. Our overview of AI tools for research describes what these tools can help with and where their limits are.

Ethics and misconduct

Good peer review rests on confidentiality, fairness and declared conflicts of interest. The Committee on Publication Ethics (COPE) publishes guidance that many journals follow, including on reviewer misconduct and fake reviews. Journals that promise publication within days, with little or no real review, are a warning sign; see how to spot predatory journals.

Peer review in open-access journals

Open access describes access, not the review process. Reputable open-access journals use the same models as subscription journals. Check the journal’s description of its review policy, whatever its business model.

Becoming a reviewer yourself

Reviewing is part of academic life and a good way to learn how editors think. Early-career researchers often start by co-reviewing with a supervisor, with the editor’s knowledge. Keep each review constructive: summarise the paper, list major and minor points, and give concrete suggestions. Write the review you would like to receive.