The ability of physostigmine alone and in combination with neostigmine to reverse d-tubocurarine-induced neuromuscular block was evaluated in surgical patients. The relaxation was maintained at a level of 90% twitch suppression during balanced anesthesia, and antagonism was attempted with physostigmine 1.5 mg x 3; neostigmine 0.5 mg x 3; neostigmine 1.0 mg x 3; or with a combination of physostigmine 0.75 mg and neostigmine 0.5 mg x 3. The measured parameters included the twitch force or EMG amplitude of the adductor pollicis brevis muscle after supramaximal 0.1 Hz stimulation and fading of these responses after repetitive 2 and 50 Hz stimuli. Although the restitution rate of twitch height and EMG amplitude were essentially the same with both antagonists, there was a considerable time-lag in regeneration of the fades after repetitive stimuli with physostigmine as compared with the neostigmine group. The addition of physostigmine to a subeffective dose of neostigmine resulted in antagonism comparable to that seen in other groups. The clinical antagonism was satisfactory in all patients receiving physostigmine. The divergence of relaxation-indicating parameters (twitch responses and fades) after physostigmine suggests dissimilar modes of action of two antagonists at the neuromuscular junction.