Patients with haematological malignancies requiring an antifungal therapy were randomly assigned to receive amphotericin B diluted in either 5% dextrose or in fat emulsion (Intralipid). Twenty-one patients were included in each group. Mean duration of amphotericin B therapy was 8.4 days in the dextrose group and 12.8 days in the Intralipid group. Amphotericin B infusion induced chills in 16 of 21 patients in the dextrose group and in 5 of 21 in the Intralipid group (P = 0.0008). Serum creatinine increased > 75% from baseline in ten patients in the dextrose group compared with only two in the Intralipid group (P = 0.007). A > or = 50% decrease of creatinine clearance was observed in 14 of 21 patients in the dextrose group compared with seven of 21 patients in the Intralipid group (P = 0.025). No difference was found between the two groups with regard to potassium and sodium requirement. Among patients who did not receive magnesium before antifungal therapy, magnesium supplementation was required more frequently in the dextrose group (8/12 vs 2/11; P = 0.02). Concomitant amikacin dosage reduction was more frequent in the dextrose group due to nephrotoxicity (7/19 vs 2/20; P = 0.045). A similar difference in vancomycin dosage reduction was observed between the two groups (12/20 vs 5/19; P = 0.03).