A multicentric randomized trial was undertaken to compare the toxicity of amphotericin B in 5% dextrose with that of amphotericin B in a fat emulsion (Intralipid) in cancer patients. Group 1 (n = 33) received amphotericin B diluted in 5% dextrose with premedication consisting of promethazine plus an antipyretic. Group 2 (n = 28) received amphotericin B diluted in 20% Intralipid without premedication. Amphotericin B was infused daily at a dose of 1 mg/kg of body weight over a 1-h period to members of both groups for empirical antifungal therapy (in neutropenic patients) or for the treatment of documented fungal infections. The majority of patients (80%) received empirical amphotericin B treatment. The two groups were comparable with regard to age, gender, underlying disease, and the following baseline characteristics: use of other nephrotoxic drugs and serum levels of potassium and creatinine. The median cumulative doses of amphotericin B were 240 mg in group 1 and 245 mg in group 2 (P = 0.73). Acute adverse events occurred in 88% of patients in group 1 and in 71% of those in group 2 (P = 0.11). Forty percent of the infusions in group 1 were associated with fever, compared to 23% in group 2 (P < 0.0001). In addition, patients in group 2 required less meperidine for the control of acute adverse events (P = 0.008), and fewer members of this group presented with hypokalemia (P = 0.004) or rigors (P < 0.0001). There was no difference in the proportions of patients with nephrotoxicity (P = 0.44). The success rates of empirical antifungal treatment were similar in the two groups (P = 0.9). Amphotericin B diluted in a lipid emulsion seems to be associated with a smaller number of acute adverse events and fewer cases of hypokalemia than amphotericin B diluted in 5% dextrose.