Steady-state plasma concentrations of paroxetine were studied at five or more paroxetine dose levels (10 to 70 mg/day) in each of 13 extensive metabolizers of sparteine and at three or four dose levels (10 to 40 mg/day) in each of three poor metabolizers of sparteine, all treated for diabetic neuropathy symptoms. On a dose of 30 mg/day there was a 25-fold variation in steady-state concentrations (25 to 670 nmol/L). The upper extreme of this variation was made up by the poor metabolizers of sparteine and the lower extreme by some fast extensive metabolizers. Further, within the extensive metabolizer group, steady-state levels showed a significant, positive correlation with sparteine metabolic ratio at all dose levels. On increasing doses, a disproportionate increase in plasma drug levels was observed in the majority of patients. In nearly all extensive metabolizers the concentration-dose data were best described by a pharmacokinetic model assuming elimination by at least two kinetically distinct processes, one a high-affinity saturable process and one a low-affinity linear process. Estimates of clearance at low drug levels of the high-affinity process showed a significant negative correlation with the sparteine metabolic ratio. Clearance of the low-affinity process was not related to the metabolic ratio and was of the same magnitude in extensive and poor metabolizers. The data thus confirmed that the metabolism of paroxetine and sparteine cosegregates and indicated that the enzyme responsible for a high-affinity saturable paroxetine elimination process is identical with CYP2D6, the source of the sparteine oxidation polymorphism.