Preterm infants are hypogammaglobulinemic at birth, a condition that worsens during the first several weeks of life. It has been postulated that periodic infusions of intravenous immune globulin in preterm infants might prevent systemic infections after age 7 days, but clinical trials have been inconclusive. To test this hypothesis in neonates weighing 500 to 1750 g at birth, a multicenter, randomized, double-blind, placebo-controlled trial was initiated. Either intravenous immune globulin (500 mg/kg) (284 infants) or placebo (5% albumin-normal saline, 10 ml/kg) (293 infants) was infused periodically for 8 weeks. Infusions were well tolerated. Mortality (4%) was similar in both groups. However, the incidence of systemic infection was significantly reduced in the treatment group, but only in those weighing less than 1500 g. More than 80% of infections were bacterial; half were caused by staphylococci. Standard use of intravenous immune globulin in preterm infants looks promising but must await full analysis of data from this and another ongoing multicenter trial.