The pharmacokinetics and pharmacodynamics of pancuronium were studied following intravenous infusion in eleven patients undergoing surgical anaesthesia. Measurement of the plasma concentrations (Cp) of the neuromuscular blocking agent ( NMBA ) and the concomitant intensities of paralysis allowed their simultaneous modelling. The pharmacokinetic parameters derived for pancuronium were in the range of previously reported values, except that the mean total systemic plasma clearance (0.79 +/- 0.28 ml X min-1 X kg-1) was reduced and the mean terminal phase half-life (169 min) was longer in these patients. Plasma concentration and % paralysis data were successfully fitted to a previously proposed pharmacodynamic model. This model assumes a separate effect compartment which exchanges drug directly with the central kinetic compartment (integrated effect model). The 'steady-state' Cp necessary to produce 50% paralysis ( ECpss (50] was estimated to be 0.21 +/- 0.08 micrograms X ml-1 (mechanical response) and 0.18 +/- 0.05 micrograms X ml-1 (EMG response). An analysis using the Hill equation of the Cp-response relationship, during and after the constant-rate infusion of pancuronium bromide, resulted in effective plasma concentrations for 50% paralysis ( ECp50 ) of 0.35 +/- 0.06 micrograms X ml-1 and 0.20 +/- 0.09 micrograms X ml-1, respectively, for mechanical twitch response. The corresponding values for EMG response were 0.32 +/- 0.06 micrograms X ml-1 and 0.17 +/- 0.06 micrograms X ml-1. Using this latter approach, the ECp50 estimated during onset of paralysis was significantly higher than that estimated during offset of paralysis (p less than 0.05); no such difference was apparent between this latter parameter and the ECpss (50) of the integrated effect model (p greater than 0.05). No significant differences were observed between any of the pharmacodynamic parameter estimates generated from the data obtained from the two methods of assessment of neuromuscular function (mechanical vs. EMG response) (p greater than 0.05).