A prospective, randomized, controlled, double-blind study was performed between 1982 and 1985 to assess the ability of ketoconazole to prevent fungal infections in selected patients with cancer. Fifty-six patients receiving induction chemotherapy for acute leukemia, autologous bone marrow transplant for refractory nonhematopoietic malignant neoplasms, multidrug chemotherapy for malignant lymphoma, or corticosteroids for brain metastases were randomized to receive either oral ketoconazole, 400 mg/d, or placebo and observed until leukopenia resolved or corticosteroid therapy was stopped. Oral candidiasis developed in eight (28%) of 29 patients receiving placebo compared with none of 27 receiving ketoconazole. However, ketoconazole failed to prevent Candida esophagitis and vulvovaginitis in two patients and one patient, respectively. Furthermore, prophylactic use of ketoconazole did not significantly alter the total number of hospital days, febrile days, or antibiotic days or the requirement for amphotericin B in patients with acute leukemia and autologous bone marrow transplant. Since oral candidiasis can be successfully managed by several different treatment modalities when it does occur, we do not think that the routine prophylactic use of ketoconazole is justified.