Platelet antibodies either bound to the surface of platelets or free in the serum were sought in patients who had low platelet counts for a variety of reasons. They were detected by finding excess IgG on the surface of washed platelets either directly or after incubation of the serum with normal platelets. The technique used was a modification of that described recently (Dixon et al, 1975) in which the greater the amount of anti-IgG consumed by the reaction with platelets the less the subsequent lysis of sheep red cells coated with IgG. This test could be calibrated by adding known quantities of IgG to the antisera and thus the amount of bound IgG could be measured. Platelets from normal donors and those with thrombocytopenia due to non-immunological causes such as aplastic anaemia or acute leukaemia were found to have 15-70 ng IgG/10(7) platelets (mean 53 ng). 37 out of 38 thrombocytopenic patients in whom immune destruction of platelets was suspected were found to have excess IgG on their platelets ranging from 70 to 720 ng/10(7) (mean 297 ng, P less than 0.001) and there was a significant inverse correlation between this amount and the platelet count (r = 0.85, P less than 0.001). Antibody in the serum was found in 14 of 22 patients with 'idiopathic' thrombocytopenic purpura (ITP), three of four patients with underlying lymphoma and in all five cases of systemic lupus erythematosus (SLE). Four non-thrombocytopenic patients with autoimmune haemolytic anaemia (AIHA) due to IgG on the red cells were also studied and were shown to have no increase in platelet-bound IgG. Our results confirm the work of Dixon et al (1975) that platelet antibody as excess IgG can be readily detected on the surface of platelets in patients with immune thrombocytopenia. The clinical implications of these findings are discussed.