We undertook a randomized, placebo-controlled, double blind trial of cytomegalovirus (CMV) immunoglobulin (CMVIG) for prevention of CMV-associated disease in 183 multiply transfused, premature neonates. CMVIG (150 mg/kg) or placebo was given within 24 hours of the first transfusion and at Day 10. If an intravenous catheter was still in place an additional dose was given between Days 20 and 30. The globulin and placebo groups were well-matched with respect to birth weight, gestational age, Apgar score, birth to a CMV-seropositive mother, requirement for assisted ventilation and exposure to CMV-positive, unscreened blood products. Among infants followed for more than 10 days, 18 (10.5%) developed CMV infection; 9 had symptomatic CMV disease (5 placebo; 4 CMVIG). Among infants born to a CMV-seropositive mother, CMVIG use was associated with a CMV syndrome rate of 3.2% (95% confidence interval, 0.2 to 18.5%) compared to 12.5% (95% confidence interval, 4.5 to 27.6%) among placebo recipients (P= 0.163). Among placebo recipients infants born to CMV-seropositive mothers were more likely to have a virologically confirmed CMV syndrome than those born to a CMV-seronegative mother, despite receipt of blood not screened for CMV antibody (P= 0.012). Multivariate analysis demonstrated that two factors were independently associated with CMV acquisition: the volume of CMV-seropositive blood products transfused (P= 0.005); and birth to a CMV-seropositive mother (P= 0.006). Infusions of CMVIG were well-tolerated. This study reaffirms that perinatally acquired CMV disease is more common among infants born to CMV-seropositive mothers than CMV-seronegative mothers, even without use of CMV-screened blood products.