Usually the researchers obtain the skills to review scientific papers not from instructions but rather from experience. There are no formal guidelines to assist referees. Reviewers are selected by editors based on their expertise, position availability and the availability of the reviewers. Being invited to review a manuscript is an honor, not only because you are being recognized for your eminence in a particular area of research but also because of the responsibility and service you provide to the journal and scientific community.

The skills needed to review scientific manuscripts develop with experience. The techniques of peer-review can always be improved and nurtured through constant efforts. Reviewers, for the most part, act in this capacity from a sense of duty, selflessness, and a desire to contribute in an important way to the maintenance of high standards and veracity in their specific areas of research. Usually, no monetary compensation is, or should be, provided.

The responsibilities of a reviewer can be summarized as follows.

1. The reviewer should provide an honest, critical assessment of the research. The reviewer’s job is to analyze the strengths and weaknesses of the research, provide suggestions for improvement, and clearly state what must be done to raise the level of enthusiasm for the work. The reviewer should not manipulate the process to force the authors to address issues interesting or important to the reviewer but peripheral to the objective(s) of the study.
2. The reviewer should maintain confidentiality about the existence and substance of the manuscript. It is not appropriate to share the manuscript or to discuss it in detail with others or even to reveal the existence of the submission before publication. There are some exceptions, if approved by the editor. One exception is that the reviewer may want a junior colleague to have the experience of reviewing and therefore may ask him/her to collaborate on a review. However, if this is done, your collaborator on the review should also agree to maintain confidentiality, and the editor should be informed of the participation of this additional person. Some journals require editor approval before a colleague or student is asked to view a submitted paper; others do not.
3. The reviewer must not participate in plagiarism. It is obviously a very serious transgression to take data or novel concepts from a paper to advance your own work before the manuscript is published.
4. The reviewer should always avoid, or disclose, any conflicts of interest. For example, the reviewer should decline to review a manuscript on a subject in which he/she is involved in a contentious dispute and does not feel that a fair review can be provided. The reviewer should also avoid biases that influence the scientific basis for a review. One example of this is a bias that favors studies with positive rather than negative results. Another example is if the reviewer has a close personal or professional relationship with one or more of the authors such that his/her objectivity would be compromised. Scientific merit should be the basis for all reviews.
5. The reviewer should accept manuscripts for review only in his/her areas of expertise. Although editors try very hard to match manuscripts with the most expert reviewers, sometimes mistakes are made. It is unfair to the authors and to the overall review process if the referee does not have the expertise to review the manuscript adequately. The exception to this general rule is when an editor specifically asks for your view as a "non-expert" or seeks your opinion on a special aspect of the manuscript (e.g., statistics).
6. The reviewer should agree to review only those manuscripts that can be completed on time. Sometimes, unforeseen circumstances arise that preclude a reviewer from meeting a deadline, but in these instances the reviewer should immediately contact the editor. It is unfair to the authors of the manuscript for reviews to be inordinately delayed by tardy referees. Delaying a review can sometimes lead to charges by the authors that the reviewers (who undoubtedly work in the same area) are "stonewallng" in order to publish their related work first, thus establishing priority.
7. The reviewer also has the unpleasant responsibility of reporting suspected duplicate publication, fraud, plagiarism, or ethical concerns about the use of animals or humans in the research being reported.
8. The reviewer should write reviews in a collegial, constructive manner. This is especially helpful to new investigators. There is nothing more discouraging to a new investigator (or even to a more seasoned one) than to receive a sarcastic, destructive review. Editors are not trying to determine the scientific prowess or wittiness of the reviewer. The reviewer should not shy away from discussing the weaknesses (or strengths) of a study, however. No one likes to have a paper rejected, but a carefully worded review with appropriate suggestions for revision can be very helpful. In fact, an author should prefer to have his/her paper rejected if the review process uncovered errors in the study.