Fifty-three children, aged 7 months to 14.4 years and with typical acute immune thrombocytopenic purpura and platelet counts < or = 20 10(9)/L, were randomly assigned to receive intravenously administered immune globulin G (IVIG), 1 gm/kg per day for 2 consecutive days (n = 19); orally administered prednisone, starting at a dose of 4 mg/kg per day, with tapering and discontinuation of corticosteroids by day 21 (n = 18); or no therapy (n = 16). Both IVIG and prednisone resulted in significantly fewer days with platelet counts < or = 20 x 10(9)/L in comparison with no therapy (median, 1 and 2 days vs 4 days; corresponding ranges, 1 to 20 and 1 to 11 days vs 1 to 132 days; p < 0.01). Reversal of clinically important thrombocytopenia assessed by the number of days taken to achieve a platelet count of > or = 50 x 10(9)/L was significantly faster in children randomly assigned to receive IVIG (median, 2 days; range, 1 to 34 days) than in those receiving prednisone (median, 4 days; range, 2 to 13 days; p < 0.001) or no therapy (median, 16 days; range, 2 to 132 days; p < 0.001). Because the risk of intracranial hemorrhage in children with acute immune thrombocytopenic purpura is highest in the group with severe thrombocytopenia, and appears to be restricted to children with platelet counts < or = 20 x 10(9)/L, these results support the use of IVIG or high doses of prednisone as initial therapy in children with acute immune thrombocytopenic purpura and severe thrombocytopenia (platelet counts < or = 20 x 10(9)/L).