Peer review. We depend on it for validation of research results. We despise it when it fails us. Most research can benefit from a serious critique. Good peer review provides constructive criticism from expert colleagues, improving the quality of research and selecting the most reliable and interesting data for publication.

How do we implement peer review at Nature Immunology? Unlike most immunology journals, Nature Immunology and the other Nature journals do not have editorial boards. This editorial freedom guarantees that no set group of individuals exercises undue sway over decision-making. In-house editors are solely responsible for maintaining the quality of peer review. We select the manuscripts most likely to be of interest to our readers and to pass review. That being said, we often consult outside experts about new submissions, as in-house expertise cannot possibly encompass all of immunology. Deciding which papers to send to review usually takes about two days.

Less than half of our submissions are sent on to in-depth review. If the question being investigated is an important one, the experimental approach suitable and the data interesting and generalizable, the paper stands a good chance of being reviewed. We ask that authors suggest at least four potential referees (specifying areas of expertise), as well as steer us from those who may give a biased critique. Nature Immunology will not pick referees exclusively from author suggestions, but we do honor all reasonable requests to avoid certain parties. Referees are also selected from our database of over 500 reviewers and from the recommendations of other referees. Finding the right person is critical; half of the papers have a full complement (usually two or three) in less than five days, but occasionally it has taken almost two weeks.

We expect much from our referees, as they are the lynchpins of peer review. When contacted, referees initially determine that there are no insurmountable personal, scientific or financial conflicts that could color their evaluation of the submitted paper. If there are any doubts, we encourage discussion with an editor. Our peer-review process is confidential. Only the referee who agreed to review the paper should read it, unless permission was received to enlist another's aid in reviewing the work. As with all confidential peer review, referees should keep information garnered from the paper confidential and not use it to further their own work. The identity of referees remains confidential even after a decision has been made on the fate of the paper.

We generally give referees two weeks to return reviews to us and understand that it can take up to half a day to prepare thoughtful analyses. Forty percent of reviews are returned before they are due, with half arriving at Nature Immunology by their due date. Re-ferees are expected to thoroughly examine the methodology, approach and implications of the data and provide some insight into the scope, significance and general interest of the conclusions. Ideally, a review succinctly summarizes the main points of the paper, then delineates where the data is not robust enough to support the conclusions. The review should conclude with concrete suggestions for improvements and a recommended course of action for the editor. Although a considered review is time-consuming, with no immediate reward for the referee, constructive critiques are the backbone of peer review—a system that benefits all investigators. Carelessly phrased or insufficiently detailed reviews can have severe consequences for authors.

Editors then evaluate all information on hand about the manuscript. Do the reviews accurately reflect the content of the manuscript? Are the comments of the referees valid? Do the differences in the referee's comments contradict one another, a sign of a potential problem, or are the differences due to referees' emphasis on different aspects of the paper? A final decision on the fate of the manuscript, then, is not just a verbatim repeat of the referees' advice, but includes an evaluation of the referees' critique. Problems occasionally arise when an author perceives that reviews were carelessly prepared and the editorial decision based on unfair or inaccurate information. If a re-examination of the file by the editors reveals a good possibility of bias or major errors, the authors will likely be offered a re-review. Ideally, this would always be recognized and remedied before the authors receive the reviews.

Thus, the review process is a fragile balance between available time and quality of effort. Editors should only review those papers that have a chance of success; reviewers should only agree to review if they have the time to do a thoughtful and constructive job; and authors should not appeal negative decisions in a frivolous manner, as resubmission requests result in time-consuming re-examinations on the part of editors and referees.

Peer review is only as good as the community it serves. Nature Immunology is constantly evaluating its practices. If you have suggestions on how to improve peer review, please write to us at immunology@natureny.com.