The ability of edrophonium and neostigmine to antagonize nondepolarizing neuromuscular blockade produced by steady-state infusions of atracurium, pancuronium, and vecuronium was studied in 71 adult patients anesthetized with nitrous oxide and halothane. Infusion rates of blocking drugs were adjusted so that single twitch depression as measured by the evoked integrated EMG of the hypothenar muscles was kept at 10% of control. Two minutes after the termination of the infusion either edrophonium (0.75 mg/kg) or neostigmine (0.05 mg/kg) was administered. Single twitch depression and train-of-four (T4/T1) fade was recorded during the recovery period. T4/T1 fade ratios observed at 20 min postreversal were 0.80 (atracurium-edrophonium); 0.76 (vecuronium-edrophonium); 0.44 (pancuronium-edrophonium); 0.95 (atracuriumneostigmine); 0.89 (vecuronium-neostigmine); and 0.68 (pancuronium- neostigmine). Under conditions of this study neostigmine produced more rapid and complete recovery than did edrophonium. Although edrophonium produced adequate antagonism of atracurium if 20–30 min were allowed to elapse, edrophonium reversal of pancuronium was rarely acceptable even at 30 min. Increasing the dose of edrophonium to 1.0 mg/kg produced single twitch values of 0.90 at 5 min postreversal but did not increase the rate of recovery of the train-of-four fade ratio. Neostigmine reversal of pancuronium, on the other hand, generally produced T4/T1 ratios of >0.70 in 20–30 min. Although the pattern of recovery seen after reversal of vecuronium was in general quite similar to that seen after atracurium, two patients in the vecuronium-edrophonium group showed delayed recovery and also failed to respond significantly to subsequent doses of neostigmine. Following steady-state infusions of vecuronium, it appears that marked patient variability in the speed of recovery can occur. Our results do not confirm other published reports that suggest that edrophonium and neostigmine may be used interchangeably.