The clinical factors influencing the results of granulocyte transfusion were evaluated, and the incidence of reactions following transfusion of granulocytes obtained by filtration and intermittent centrifugation leukapheresis was compared. Thirty-seven patients received 195 evaluable transfusions (99 filtration and 96 centrifugation) for 45 infectious episodes. The mean yields were 22.2 × 109 granulocytes for filtration and 8.3 × 109 granulocytes for centrifugation using non-stimulated donors. Prospectively, 58 transfusions (29 filtration and 29 centrifugation) were transfused to 17 matched donor—recipient pairs and 12 (40%) moderate or severe reactions occurred after filtration transfusions while two (7%) moderate reactions occurred after centrifugation transfusions. Overall 36% of filtration transfusions and 3 % of centrifugation transfusions were associated with reactions. Among 18 infectious episodes which improved, five patients improved only because of granulocyte transfusions. Among 11 episodes which stabilized, six stabilized only because of transfusions. Sixteen episodes progressed. The five patients who improved and the six who stabilized only because of transfusions, received more granulocytes than those who progressed. Responses were seen following transfusion of granulocytes collected by either filtration along or both methods. Only three patients clearly progressed and they received centrifugation granulocytes alone suggesting that the dose of granulocytes derived from intermittent centrifugation may have been suboptimal.