A randomized clinical trial to determine the efficacy of granulocyte transfusions in neutropenic patients with infection was conducted. Criteria for patient selection included a proved infection, a granulocyte count of less than 300/mm 3, availability of a suitable donor and failure to respond to at least 72 hours of appropriate antibiotic therapy. Thirty patients were assigned at random to receive either granulocyte transfusions or to serve as a control group. Antibiotic therapy was continued in both groups. Responses were judged by the degree of diminution of infectious episodes and survival. The results showed that 11 of 13 control patients failed to respond during the period of observation, whereas 10 of 17 patients given transfusions responded. The results were statistically significantly different (p < 0.05). The median survival was 22.5 days in the group given transfusions (group 2) and 7.7 in the control group (group 1) (p < 0.01). The granulocyte transfusions were most effective in patients with hypocellular marrows who failed to recover during the period of observation. These results indicate that granulocyte transfusions are effective in the short-term control of infections in neutropenic patients.