The efficacies of antifungal prophylaxis with ketoconazole and itraconazole, a new triazole, in patients with prolonged granulocytopenia were evaluated in two nonrandomized studies. The conditions other than the drug administered for prophylaxis were equivalent in the two studies. The incidence of fatal fungal infections was significantly higher among patients given ketoconazole than among those given itraconazole (P = .02); this trend was especially evident with fatal infections due to Aspergillus (P = .0045). The level of protection from fatal fungal infection afforded by itraconazole was highly significant among patients who were granulocytopenic for more than 25 days (P less than .0001) and among patients with acute lymphoblastic leukemia (P = .008). Failures of prophylaxis with itraconazole may have been due to inadequate levels of drug in plasma. Although these results must be interpreted with caution because they were obtained in consecutive nonrandomized studies, the use of itraconazole for antifungal prophylaxis in granulocytopenic patients seems to be a major advance and to be especially advantageous in hospitals, where the incidence of fatal aspergillosis in these patients is high.