To determine the potencies of neostigmine, pyridostigmine, and edrophonium in reversing pancuronium and d-tubocurarine blockade, dose-response curves were established for first twitch height recovery and train-of-four ratio. One hundred and twenty ASA physical status I or II patients scheduled for elective surgery received either 0.06 mg/kg pancuronium or 0.36 mg/kg d-tubocurarine during a thiopental-nitrous oxide-enflurane anesthetic. Train-of-four stimulation was applied every 12 s, and the force of contraction of the adductor pollicis muscle was recorded. When first twitch height had recovered spontaneously to 10% of its initial value, neostigmine (0.005, 0.01, 0.02 or 0.05 mg/kg), pyridostigmine (0.02, 0.04, 0.1, or 0.2 mg/kg), or edrophonium (0.1, 0.2, 0.4 or 1 mg/kg) was injected by random allocation. Recovery was measured 10 min after the injection of the antagonist. First twitch ED50's were 0.013, 0.085, and 0.17 mg/kg after pancuronium, and 0.017, 0.11, and 0.27 mg/kg after d-tubocurarine, for neostigmine, pyridostigmine, and edrophonium, respectively. The ED50 for pyridostigmine and edrophonium obtained after d-tubocurarine was significantly larger (P less than 0.05) than that after pancuronium. The train-of-four dose-response curves were significantly flatter for edrophonium than for the other two agents, indicating a greater ability of edrophonium to antagonize fade at low doses. It is concluded that the potency of reversal agents may be different for different relaxants, and that potency ratios might depend upon the end-point chosen as full neuromuscular recovery.