OBEDIENCE serves numerous productive functions in society. It may be ennobling and educative and entail acts of charity and kindness. Yet the problem of destructive obedience, because it is the most disturbing expression of in our time, and because it is the most perplexing, merits intensive study. In its most general terms, the problem of destructive may be defined thus: If X tells Y to hurt Z, under what conditions will Y carry out the command of X, and under what conditions will he refuse? In the concrete setting of a laboratory, the question may assume this form: If an experimenter tells a subject to act against another person, under what conditions will the subject go along with the instruction, and under what conditions will he refuse to obey? A simple procedure was devised for studying (Milgram, 1963). A person comes to the laboratory, and in the context of a learning experiment, he is told to give increasingly severe electric shocks to another person. (The other person is an actor, who does not really receive any shocks.) The experimenter tells the subject to continue stepping up the shock level, even to the point of reaching the level marked Danger: Severe Shock. The purpose of the experiment is to see how far the naive subject will proceed before he refuses to comply with the experimenter's instructions. Behavior prior to this rupture is considered obedience in that the subject does what the experimenter tells him to do. The point of rupture is the act of disobedience. Once the basic procedure is established, it becomes possible to vary conditions of the experiment, to learn under what circumstances to authority is most probable, and under what conditions defiance is brought to the fore (Milgram, in press). The results of the experiment (Milgram, 1963) showed, first, that it is more difficult for many people to defy the experimenter's authority than was generally supposed. A substantial number of subjects go through to the end of the shock board. The second finding is that the situation often places a person in considerable conflict. In the course of the experiment, subjects fidget, sweat, and sometimes break out into nervous fits of laughter. On the one hand, subjects want to aid the experimenter; and on the other hand, they do not want to shock the learner. The conflict is expressed in nervous reactions. In a recent issue of American Psychologist, Diana Baumrind (1964) raised a number of questions concerning the report. Baumrind expressed concern for the welfare of subjects who served in the experiment, and wondered whether adequate measures were taken to protect the participants. She also questioned the adequacy of the experimental design. Patently, Behavioral Study of Obedience did not contain all the information needed for an assessment of the experiment. But it is clearly indicated in the references and footnotes (pp. 373, 378) that this was only one of a series of reports on the experimental program, and Baumrind's article was deficient in information that could have been obtained easily. I thank the editor for allotting space in this journal to review this information, to amplify it, and to discuss some of the issues touched on by Baumrind. At the outset, Baumrind confuses the unanticipated outcome of an experiment with its basic procedure. She writes, for example, as if the production of stress in our subjects was an intended and deliberate effect of the experimental manipulation. There are many laboratory procedures specifically designed to create stress (Lazarus, 1964), but the paradigm was not one of them. The extreme tension induced in some subjects was unexpected. Before conducting the experiment, the procedures were discussed with many colleagues, and none anticipated the reactions that subsequently took place. Foreknowledge of results can never be the invariable accompaniment of an experimental probe. Understanding grows because we examine situa-