To test the efficacy of poststorage bedside leucodepletion of blood products in the prevention of primary HLA alloimmunization and its clinical sequelae, 172 patients with hematologic malignancy requiring intensive red blood cell and platelet support were randomized to receive either standard or filtered red blood cells and platelets. Quality control of bedside filtration was explored by sequential sampling downstream of the filter, but this did not predict the total number of leucocytes transfused. After exclusions, 123 evaluable patients were assessed every two weeks until the end of therapy. HLA antibodies developed in 21 of 56 (37.5%) nonfilter (NF) and 15 of 67 (22%) filter (F) patients (risk ratio estimate, 0.60 [95% confidence interval, 0.34 to 1.05]; P = .07). Patients with acute myeloid leukemia (AML; n = 53) had higher alloimmunization rates in both arms of the study, with a greater effect of filtration (62.5% NF and 31.0% F; P = .025). Bedside filtration did not affect the overall incidence of febrile transfusion reactions (FTRs; 37% NF and 34% F; P = .71) or of platelet refractoriness assessed in 50 patients (30% NF and 26% F), despite an association between broad HLA reactivity and both FTRs and refractoriness. However, FTRs were also seen in 28 patients without HLA antibodies. Five alloimmunized refractory patients (2 F and 3 NF) required HLA-selected platelets. This report, the first prospective study of bedside filtration, has failed to show clear clinical benefit. Methodological limitations may account in part for this failure, notably the difficulties in accurately assessing the number of leucocytes transfused.