The pharmacokinetic-pharmacodynamic (PK-PD) relationship of vecuronium administered as a 0.1 mg/kg intravenous bolus was evaluated during both the onset of and recovery from neuromuscular block in 9 anaesthetised patients. Sigmoid Emax modelling of neuromuscular block versus vecuronium effect compartment concentrations indicated comparable effective concentrations at 50% block (EC50) for both onset and recovery (146 ± 12 vs 144 ± 13 ng/ml, respectively). However, during onset, the sigmoid was systematically steeper (gamma of 6.41 ± 0.32 vs 4.17 ± 0.36, respectively). Including a finite receptor concentration in the effect compartment (0.27 ± 0.04 μmol/L) led to a faster equilibration rate constant, Keo (0.097 ± 0.009 vs 0.058 ± 0.005 min-1) and to lower EC50 (120 ± 13 ng/ml) and gamma values (3.62 ± 0.22). The neuromuscular block predicted by this model was similar to the one obtained from a separate PK-PD analysis over each phase. Therefore, to be adequate, the PK-PD modelling of non-depolarising neuromuscular blockers should involve a separate analysis of onset or recovery data, or account for the safety margin of neuromuscular transmission by adding a receptor concentration in the effect compartment.