In a randomised, multicentre study intravenous IgG was compared with oral corticosteroids in 108 children with untreated acute immune thrombocytopenic purpura. IgG was an efficient treatment with no severe untoward reactions. The effects of corticosteroids and IgG were identical for rapid responders, who accounted for 62% of all patients. In contrast, patients requiring more than initial treatment responded better if randomised to IgG. The serum IgG level increased two-fold after IgG. A significant rise in IgM levels was observed after both IgG and corticosteroids. The platelet-associated IgG index was high in 75% of all patients. No significant differences between the two treatment groups were found, but rapid responders had a smaller mean initial platelet-associated IgG index which returned more rapidly and more permanently to normal than that of slow responders.