A B S T R A C T To study the antibody response to human platelet transfusions, nine thrombocytopenic patients with bone marrow failure were given 6 U (3 X 10") of random platelet concentrates twice a week.Before transfusion, none of the patients had preexisting antibodies detectable with lymphocytotoxicity, platelet aggregation, or capillary leukoagglutination techniques. After receiving 18-78 U of platelets, they became refractory to further transfusions of random platelets and alloantibodies were detectable. Two patterns of antibody response could be identified. In three patients, the sera were not lymphocytotoxic with a panel of standard cells in which all the known HLA antigens in the first and second series were represented at least once. Yet, they caused platelet aggregation with 30, 24, and 60%, respectively, of a donor population studied. The aggregating activities were inhibited by antihuman IgG but not by antihuman IgA or antihuman IgM antiserum. The aggregating antibodies could be absorbed out with donor platelets but not lymphocytes or granulocytes. Antibodies from two of these patients aggregated platelets of their respective siblings matched for both HLA haplotypes. Transfusion of platelets from these two siblings did not increase the platelet count while platelets obtained from aggregation-negative donors did. The sera from the remaining six patients were lymphocytotoxic with 15-100% of the panel of standard cells. They also had aggregating antibodies, which could be absorbed out by both platelets and lymphocytes, suggesting that they were HLA antibodies. These data suggest that the development of platelet-specific antibodies may play an Dr. Wu's current address is Section