OBJECTIVE In patients with severe head injuries receiving long-term infusion for reducing intracranial pressure, a decline in concentrations was apparent following attainment of an initial steady state. This could be explained by an increased rate of elimination. An adequate modeling of the plasma disposition curves was used to demonstrate clearly the metabolic induction. METHODS The concentration-time data of 17 patients were fit by a one compartment pharmacokinetic model in which the decline of plasma concentration during infusion was due to an increase in the clearance rate of thiopental following a latency period. This time-dependent clearance model provided estimates of initial and final clearance rates. RESULTS This study demonstrated that large interindividual variations were observed during the course of the thiopental time-dependent pharmacokinetics. Depending on the patient, one or two steps of induction occurred. The mean initial and final clearance rates were 1.22 +/- 0.82 mL/min/kg and 10.5 +/- 23 mL/min/kg. The latency period for the first induction averaged 69 +/- 56 h. For 6 subjects, the rate of thiopental metabolism continued to change with time and there was a second step of induction. CONCLUSIONS Induction of thiopental metabolism occur within therapeutic ranges, but it was not established that attainment of individual limits in dosing rate, total dose, or treatment duration occur in the process. Thus, monitoring is needed for achievement of a target plasma concentration.