Three models for the elimination of lidocaine in the isolated perfused rat liver were used to simulate the time course of both lidocaine and its metabolites in a single-pass perfusion system: well stirred, parallel tube, and a two-compartment model to test the effects of enzyme heterogeneity. All models included multiple enzymes and multiple metabolic pathways, as well as varying degrees of tissue binding. Although the well stirred and parallel tube models gave qualitatively different results, neither model predicted that the concentration of monoethylglycinexylidide would pass through a maximum and then decline to a lower steady state value, as observed in continuous perfusion experiments. Although each of the three models tested would give reasonable agreement with steady state observations, the test of the time-dependent behavior of both lidocaine and monoethylglycinexylidide was more discriminating. Each model gave characteristic predictions for the time course of the metabolites.