A multicenter prospective, randomized controlled trial was conducted to determine whether early use of platelet concentrates would reduce the incidence or extension of intracranial hemorrhage or both in sick preterm infants with thrombocytopenia. The effects on bleeding as reflected by the amount of blood product support administered and a shortened bleeding time were assessed as secondary outcomes. Premature infants with a platelet count < 150 x 10(9)/L within the first 72 hours of life were randomly assigned to receive either conventional therapy or conventional therapy plus platelet concentrates (10 ml/kg). The platelet count was maintained < 150 x 10(9)/L until day 7 of life by one to three platelet transfusions. In 22 (28%) of the 78 treated infants and 19 (26%) of the 74 control infants, either a new intracranial hemorrhage developed or an already-present one became more extensive (p = 0.73). Similar numbers of infants had each grade of intracranial hemorrhage on both initial and follow-up ultrasonography. Similar numbers of infants received fresh frozen plasma and packed red blood cells, but treated infants received less of both. The bleeding time was prolonged in the treated group before the infusion of platelet concentrates but subsequently shortened (mean difference, 79.0; 95% confidence interval, 73.1 to 84.9). Subanalysis of the control group showed that infants with platelet counts < 60 x 10(9)/L (n = 21) on at least one occasion received more fresh frozen plasma and packed red blood cells than did those with platelet counts > 60 x 10(9)/L.(ABSTRACT TRUNCATED AT 250 WORDS)