The effectiveness of intravenously administered immunoglobulin (Ig) therapy for prophylaxis of infection was evaluated in high-risk preterm infants. Two hundred thirty-five premature newborns were randomly assigned, in a double-blind controlled trial, to treatment and placebo groups. Thirty-five infants (29%) of the Ig group and 29 (25%) of the placebo group had one or more episodes of certain infection. Thirty infants (25%) of the Ig group and 18 (16%) of the placebo group had one or more episodes of probable infection. No significant differences were observed in the incidence of certain or probable infection in treated and control infants. Nevertheless, among the infants who had one or more certain or probable episodes of infection, more of them belonged to the Ig group than to the placebo group. The possible deleterious effect of the administration of large amounts of polyspecific Ig is discussed.