We prospectively studied the efficacy and adverse effects of chlorpromazine (30 mg/m2 given intravenously) plus lorazepam (0.04 mg/kg given intravenously) versus chlorpromazine alone in a controlled, double-blind, randomized, parallel-design investigation in 25 children (1.5 to 17.3 years of age) with acute lymphoblastic leukemia. Response to other antiemetics in eight children refusing random assignment to treatment was also evaluated. All children were receiving intravenous infusions of teniposide plus cytarabine, the pharmacokinetics of which were characterized for each of the one to four courses. There were no differences between the 11 patients randomly assigned to receive chlorpromazine alone and the 14 randomly assigned to receive lorazepam plus chlorpromazine in the number of emesis episodes (6.0 vs 5.9; p = 0.53), frequency of dystonic reactions (3% vs 5%), or akathisia (13 vs 10%). The only serious adverse event, symptomatic hypotension, occurred in a boy receiving chlorpromazine plus lorazepam. An exploratory pharmacodynamic analysis revealed that the only variable that correlated with vomiting was cytarabine 1 1/2-hour plasma concentration (p = 0.007). Children who received either chlorpromazine plus lorazepam or chlorpromazine alone had fewer episodes of vomiting than those who received "conventional" antiemetic therapy (6.0 vs 8.6; p = 0.01). We conclude that the severity of emesis is related to the plasma concentration of cytarabine; that intravenously administered chlorpromazine is as effective as chlorpromazine plus lorazepam in preventing chemotherapy-induced vomiting; and that the potential for adverse effects with the addition of lorazepam may be a disadvantage.