The authors studied an epidemic of hepatítis B occurring in an institution located in the São Cristovão section of the city of Rio de Janeiro, Brasil. Of a total of 35 cases detected, 30 ocurred in employees of the institution and 5 in family members of the employees. All the cases were considered as benign acute hepatitis; all but two were iteric. Clinically, a marked articular involvement was noted, with intense arthralgia and pronounced restriction of movement, which regressed completely as soon as jaundice became evident. The chronological distribution was typical of simultaneous infection from a single source. Epidemiological investigation indicated that the source was commercially prepared pooled human gamma globulin. Ironically, injections were given, as prophylaxis against hepatitis with disposable syringes. During two days of application, approximately 120 employees received gamma globulin injections, among whom 27 developed hepatitis with a mean incubation period of 116 days. Twenty six patients were tested for HBAg, with positive results in 12 (46.2%); 8 patients tested during the first two weeks of overdisease were all positive. The epidemic outbreak occurred in conditions, though not premeditated, approximating to those of an experimental study. In addition to the São Cristovão epidemic, the authors studied 8 cases seen at the hospital Service where they worked. These cases presented characteristics distinguishing them from the the type of hepatitis case commonly treated in that hospital. All these 8 patients were of elevated socioeconomic conditions, and all had a history of having received injections of gamma globulin from the same commercial source as that which gave rise to the São Cristovão epidemic. In these cases the mean incubation period was 106 days, and 5 of 6 patients tested were positive for HB Ag. Eleven lots of gamma globulin were tested in reference laboratories of the World Health Organization, six of these (53,5%) were positive for HB Ag. The authors conclude that there was a serious lapse in the process of preparation of the gamma globulin in question. It is recommended that the widespread assumption that this product is free of risks of transmitting hepatitis virus should be accepted with reservation.