An open, prospective, randomized pilot study was performed to assess the efficacy and safety of oral fluconazole 3 mg/kg once daily compared with oral nystatin 50,000 units/kg/day in four divided doses in preventing candida infections in 50 children undergoing remission induction or consolidation therapy for cancer. In 21 of 25 fluconazole-treated and 20 of 25 nystatin-treated patients the overall outcome of prophylaxis was clearly successful. Mild and transient oropharyngeal candidosis was observed in two and three patients in the fluconazole and nystatin groups respectively. One patient randomized to fluconazole and two patients randomized to nystatin required empirical treatment with amphotericin B and one patient assigned to fluconazole developed tissue-proven candida colitis. Initially non-colonized patients remained yeast-free throughout treatment with no differences between the two study arms. Initially colonized patients stayed colonized throughout treatment although at the end of the study, more patients randomized to nystatin were still harbouring yeasts (P = 0.05). Almost exclusively, Candida albicans (95%) was isolated. A change in species was observed in one patient in each arm of the study. Candida krusei or Candida glabrata were not encountered. Transient elevations of hepatic transaminases were more common in the fluconazole group, although not statistically significant (28% vs 12%, P = 0.15). Reversible grade I gastrointestinal and skin symptoms were observed in four patients randomized to fluconazole (16 vs 0%, P < 0.05). Fluconazole was as safe and effective as nystatin in controlling yeast colonization and in preventing superficial and invasive candida infections and the empirical use of amphotericin B in children and adolescents undergoing intensive chemotherapy for cancer.