In a short-time study, 8 patients with thrombocytopenia and clinical signs of bleeding were experimentally treated with prednisolone as the only therapeutic measure to control the bleeding tendency. 2 patients had idiopathic thrombocytopenic purpura and 6 acute non-lymphocytic leukaemia. 1 of the leukaemic patients was thrombocytopenic on admission. 5 had thrombocytopenia due to antileukaemic chemotherapy. The capillary fragility test, evaluated blindly, and bleeding time determination were performed just before treatment and repeated once after 2–4 d of treatment. Platelet counts were carried out simultaneously. Capillary resistance, determined by the capillary fragility test, improved greatly during the administration of prednisolone, irrespective of the platelet count, and no serious bleeding occurred. The bleeding time remained prolonged.