We report a prospective, randomized, controlled study of antifungal chemoprophylaxis in forty immunocompromised children with hematologic malignancies, receiving respectively itraconazole, ketoconazole, amphotericin-B and no prophylaxis. Fungal isolates from patients' saliva or stools were obtained from 19/40 patients (Candida albicans from 15 patients). Disseminated infection was never observed. Fungal isolates were significantly less frequent in the ketoconazole group of patients vs any other group. Systemic antifungal chemoprophylaxis with ketoconazole appears more effective even than the recently introduced itraconazole.