Subjects in a boys' secondary school were asked to participate in an experiment which exposed them to the risk of up to 50% hearing loss. Of 42 Ss, 39 participated voluntarily. Each operated a fake “sound generator” similar to Milgram's “shock generator” and administered to himself a supposed stimulus whose “danger level” was prominently displayed before him. The highest level of stimulus to which each was prepared to go was recorded. Results showed a close similarity with those obtained by Milgram when his Ss were instructed to administer dangerous shocks to others. The correspondence between the results in obedience experiments and Allport's J-curve of institutional conformity is discussed. It is contended that interpretations of experimental obedience as deriving from release of aggression or from the acceptance of malevolent intent as compatible with legitimate authority are inadequate. It is argued that both inside and outside the laboratory authority commonly functions to direct behavior in cases where the immediate and necessary steps to a worthwhile goal are nongratifying, and that experimental Ss concede authority to the E on this basis.