We used a statewide program for the surveillance and reporting of infections acquired in the hospital to examine the relation between volume of surgery and the incidence of postoperative wound infection. Over a 29-month period 25,941 surgical procedures performed at 22 hospitals were studied. For all procedures studied, the mean number of operations performed was directly related to hospital size (number of beds). A highly significant inverse relation was found between the logarithm of the frequency of operation and the infection rate for appendectomy, herniorrhaphy, cholecystectomy, colon resection, and abdominal hysterectomy. (The relation was borderline [P = 0.055] for laminectomy and not significant for cesarean section.) Although these data clearly demonstrate higher morbidity in hospitals performing relatively little surgery, there are several possible explanations, and no conclusions for health policy can yet be drawn.