To study the clinical efficacy of granulocytes obtained by filtration leukapheresis, patients with clinically evident infection and granulocyte counts of smaller than 500 per cubic millimeter were randomly assigned to receive conventional therapy alone or with a granulocyte transfusion obtained from a single donor each day for four days. Five of 19 control patients survived to day 20, and 15 of 17 in the transfused group survived. Comparison of the two populations for variables such as age, disease, and severity and type of infection revealed no other factor that could account for the difference in survival. Outcome was not demonstrated to be related to HL-A match, post-transfusion counts, or presence of leukocyte antibodies. Functional studies of granulocytes obtained by filtration leukapheresis showed only minor differences although appearance was altered. Granulocytes so obtained can be used safely and efficaciously as adjunctive therapy for infection associated with granulocytopenia.