Seventy-two adult surgical patients were studied to compare neuromuscular and cardiovascular effects of mivacurium chloride during nitrous oxide-fentanyl-thiopentone (BAL group) or nitrous oxide-halothane (HAL group) anaesthesia. Eighteen patients in the BAL group received an initial bolus of mivacurium, either the ED25 (n = 9) or the ED50 (n = 9) (0.03 and 0.05 mg kg-1). These doses were based on the assumption that the slope of the dose-response curve during nitrous oxide-opioid anaesthesia would be approximately the same as the slope of the neuromuscular response from the first human studies with mivacurium. Twenty-seven additional patients were allocated to subgroups of nine patients to receive mivacurium 0.04, 0.08 or 0.15 mg kg-1. Twenty-seven patients in the HAL group were allocated also to subgroups of nine patients to receive mivacurium 0.03, 0.04 or 0.15 mg kg-1. During stable anaesthesia, mean endtidal halothane concentrations were maintained at 0.49 +/- 0.01%. The estimated ED50, ED75 and ED95 for BAL and HAL groups were 0.039, 0.05 and 0.073 mg kg-1 and 0.040, 0.053 and 0.081 mg kg-1, respectively. Halothane did not potentiate maximum block or time to maximum block. Halothane did affect spontaneous recovery. With the 0.15-mg kg-1 dose, time to 95% recovery was prolonged significantly in the HAL group (30.0 (SEM 1.4) min) compared with the BAL group (24.1 (1.5) min). Recovery index from 25% to 75% recovery was also prolonged significantly in the HAL group (7.0 (0.4) min) compared with the BAL group (5.4 (0.4) min). There were no significant haemodynamic changes in groups given mivacurium doses up to and including 2 x ED95 by bolus i.v. administration.