Résumé : Thirty-six patients undergoing elective surgery were studied after obtaining their informed consent. They were randomly assigned to six series of six patients each. One hour before anaesthesia, all patients received 0.2 mg · kg-1 diazepam orally. After induction of anaesthesia with 1-1.5 mg · kg-1 methohexitone and 5 μg · kg-1 fentanyl, the patients were paralysed and ventilation was controlled manually (semi-open circuit; 50 % N2O/50 % O2). Each patient receveid a single dose of either 70 μg · kg-1 fazadinium, 70 μg · kg-1 pancuronium, 2,500 μg · kg-1 gallamine or 450 μg · kg-1 d-tubocurarine. Neuromuscular function was monitored by measuring the isometric contraction of the adductor pollicis muscle in response to supramaximal stimulations of the ulnar nerve at the wrist (square wave pulse of 0.2 ms duration at supramaximal intensity delivery at 0.1 Hz). Three parameters were measured : the time between the injection of the relaxant drug and recovery of the twitch height at 50 % of its baseline (RT0-50); the time between the injection of the relaxant drug and recovery of the twitch height at 90 % of its baseline (RT0-90); the time between the injection of the relaxant drug and recovery of the twitch height from 25 to 75 % of its baseline (RT25-75). The values of the observed parameters were expressed in minutes (x̄±sem). They were for RT0-50 : vecuronium 27±8, atracurium 34±7, fazadinium 47±9, pancuronium 58±15, gallamine 67±27, d-tubocurarine 103±40; for RT0-90 : vecuronium 36±13, atracurium 52±10, fazadinium 73±17, pancuronium 99±15, gallamine 132±30, d-tubocurarine 153±24; for RT25-75 : vecuronium 10±7, atracurium 16±4, fazadinium 25±7, pancuronium 47±7, gallamine 70±11, d-tubocurarine 91±25. A 1 to 3 ratio can be seen between the groups, despite the same maximum degree of paralysis being obtained for the different relaxant drugs used. Within a group, a great variation of 1 to 4 between the patients was shown up. The relationship between the different drugs, for the three parameters measured, was maintained whatever the parameter considered. In conclusion, it appeared clearly that for the different non depolarizing relaxants studied, even at equipotent doses, a great between- and within-group variability existed in normal patients, whatever the recovery parameters considered. It was thus shown that the administration of non-depolarizing muscle relaxant drugs must be adapted to clinical signs or, in a more objective manner, to the monitoring of the neuromuscular function. The administration of these drugs following a predetermined plan could induce quite unpredictable results, even in normal patients. © 1987 Masson, Paris.