Depletion of leukocytes from all blood products may decrease the incidence of alloimmunization to HLA antigens present on the white cells and thus delay the onset of refractoriness to random donor platelet support. In order to test this hypothesis, 54 patients with hematologic malignancy or marrow aplasia were entered on a prospective randomized trial using cotton-wool filtration as a method of leukocyte depletion of red cell and platelet concentrates. Forty patients were considered evaluable; 20 patients received filtered products and 20 patients in the control group received standard unfiltered products. The filter was 99% efficient in removal of leukocytes (average number of WBC/platelet product, 6 X 10(6)). Platelet loss by this technique was 8%. Alloimmunization was assessed by detection of de novo formed lymphocytotoxic and platelet specific antibodies by microcytotoxicity test, Staph A protein radioimmunoassay, and solid phase red cell adherence test. In the group receiving filtered products, three of 20 (15%) patients developed lymphocytotoxic antibodies while ten of 20 (50%) patients in the control group developed cytotoxic antibodies (P = .01 by actuarial methods). Platelet antibodies were detected in seven of ten alloimmunized patients in the control group and three of three patients in the study group. Clinical evidence of refractoriness was seen in three of 20 patients in the filtered group and ten of 20 in the control group (P = .01 by actuarial methods). The cost of filtration was a fraction of the cost of a plateletpheresis product. Filtration appears to be an effective and economical method for reducing alloimmunization and clinical refractoriness to random donor platelets in patient receiving long-term transfusion support.