The nonlinear mixed-effects modeling (NONMEM) computer program was used to investigate the variability in the duration of doxacurium-induced neuromuscular block in 408 patients enrolled in phase II and phase III clinical trials of doxacurium. Spontaneous recovery data in the 10% to 90% block range from all patients were pooled and fitted to a linear model. Two parameters were estimated: (1) the slope, which is related to the pharmacokinetics and to the steepness of the dose-response curve, and (2) the intercept, which is linearly related to dose but has no physiologic meaning. The primary goal was to determine the factors affecting the slope by use of univariate and multivariate analyses techniques. Estimates of the slope ranged from 0.67% to 1.1% block/min (interindividual variability, 39%). Factors with clinically significant effects on the slope included the following: age, obesity, and anesthesia type. Thus these factors influence the time course of doxacurium-induced block and may require individualization of dose.