Neostigmine 0.06 mg . kg-1 or edrophonium 1 mg . kg-1 were administered to two groups of 15 patients each for antagonism of pipecuronium-induced neuromuscular block at 20% spontaneous recovery of the first twitch (T1) of the train-of-four (TOF) stimulation. The mean onset of action (+/- SEM) of edrophonium (18.1 +/- 2.4 sec) was significantly more rapid (P < 0.01) than that of neostigmine (47.6 +/- 4 sec), as were the times taken to attain a TOF ratio of 0.25 and 0.5. Nevertheless, the reversal time (time taken from the end of injection of the antagonist until TOF ratio value had reached 0.75) was significantly shorter (P < 0.01) in the neostigmine than in the edrophonium group (499.3 +/- 62 vs 767 +/- 52 sec respectively). The TOF ratio ten minutes after reversal was greater in the neostigmine group than in the edrophonium group (P < 0.01), 0.78 +/- 0.02 vs 0.68 +/- 0.02 min respectively. At that time, 33% (5 out of 15) and 80% (12 out of 15) patients failed to be reversed adequately (TOF ratio of 0.75) after neostigmine 0.06 mg . kg-1 and edrophonium 1 mg . kg-1, respectively. Administration of one additional dose (one-third of the initial dose) of the same antagonist resulted in adequate antagonism in the remaining five patients in the neostigmine group and in nine patients in the edrophonium group. Two such doses were required in the remaining three patients in the latter group. The mean total dose of neostigmine and edrophonium employed in this study was 0.067 +/- 0.002 and 1.3 +/- 0.05 mg . kg-1, respectively. Under the conditions of this study, edrophonium in a dose of 1 mg . kg-1 did not consistently antagonize residual neuromuscular blockade induced by pipecuronium at 20% recovery of T1.