A prevalence of hepatitis C virus (HCV) antibodies of 12% was found in 276 patients from 11 dialysis units. Between zero and 22% of the patients in the different units were anti-HCV positive. The epidemiology of HCV was studied in two units during a 2 year period by antibody assays and the polymerase chain reaction and correlated with clinical manifestations. Two types of epidemiologic patterns were found that may explain the wide difference of HCV prevalence described in different dialysis units. In one unit there was no evidence of spread within the unit, and the prevalence of HCV was dependent on the status of the patients entering for treatment. In the other unit, a clustering of infected patients could be seen in which 13 of 36 were infected during a 3 year period. Some patients who had not received blood transfusions were among the infected. Hepatitis C infection was the most common explanation for repeated abnormal transferase levels. Most of the HCV-infected patients reacted both for anti-HCV and HCV RNA. HCV RNA was in general detected earlier than anti-HCV seroconversion. Among 20 HCV RNA-positive serum samples that were anti-HCV ELISA-positive 18 had indeterminate and two negative reactions by immunoblot (RIBA 2). Thus the RIBA 2 test should be used with caution as a confirmatory antibody test in this group of patients.