Platelets, expressing only class I antigens on their surface, have been shown to improve renal allograft survival in some experimental models but do not lead to humoral sensitization. In this pilot study, 24 previously untransfused patients were given 3 platelet transfusions at 2-week intervals before renal transplantation. Twelve patients (group 1) received 200 X 10(8) platelets contaminated by 15 X 10(6) leukocytes in each transfusion, and 42% developed cytotoxic antibodies; 12 further patients (group 2) received a purified preparation containing 200 X 10(8) platelets and less than 5 X 10(6) leukocytes, and no cytotoxic antibodies were detected. Using a more sensitive flow cytometry technique, 83% of group 1 but only 17% of group 2 (P less than 0.004) responded to their platelet donors. From each group 9 patients have been transplanted. Four grafts in group 1 and 6 in group 2 have failed, yielding a 1-year graft survival rate of 44%, compared with 69% in blood-transfused patients (P = 0.03). Pure platelet transfusions, using our protocol, do not appear to improve renal allograft survival, and minimal contamination by leukocytes can lead to humoral sensitization.