To evaluate the use of intravenously administered immune globulin (IVIG) for prevention of sepsis in preterm infants, we administered IVIG in a protocol designed to maintain a therapeutic serum "target level" of 700 mg/dl. The 200 patients who were eligible for the study (600 to 2000 gm birth weight) were monitored throughout their initial hospitalization. Of these, 115 patients were randomly assigned in a double-blind, controlled trial to treatment and placebo groups. The remaining 85 infants were not randomly assigned to a group, by parental request, but were followed and analyzed separately. In one patient who received IVIG, transient tachycardia and a decrease in blood pressure developed during an infusion; resolution occurred promptly after the infusion was discontinued. No persistent hepatic or renal abnormalities were noted in either the IVIG- or the placebo-treated group. There were seven episodes of sepsis in the placebo group and nine in the group whose parents refused consent to the study. No infant who received IVIG acquired nosocomial sepsis (p less than 0.01). All patients in the placebo group in whom sepsis developed had serum IgG levels less than 400 mg/dl at the time sepsis developed. Serum IgG levels were maintained near 700 mg/dl in patients who received IVIG. These data indicate that administration of sufficient IVIG to maintain target serum IgG levels throughout hospitalization may decrease the incidence of nosocomial sepsis in preterm infants.