Paroxetine, a selective serotonin reuptake inhibitor, is a potent inhibitor of cytochrome P450 2D6 (CYP2D6) activity, but the mechanism of inhibition is not established. To determine whether preincubation affects the inhibition of human liver microsomal dextromethorphan demethylation activity by paroxetine, we used a two-step incubation scheme in which all of the enzyme assay components, minus substrate, are preincubated with paroxetine. The kinetic parameters of inhibition were also estimated by varying the time of preincubation as well as the concentration of inhibitor. From these data, a Kitz-Wilson plot was constructed, allowing the estimation of both an apparent inactivator concentration required for half-maximal inactivation (K(I)) and the maximal rate constant of inactivation (k(INACT)) value for this interaction. Preincubation of paroxetine with human liver microsomes caused an approximately 8-fold reduction in the IC(50) value (0.34 versus 2.54 microM). Time-dependent inhibition was demonstrated with an apparent K(I) of 4.85 microM and an apparent k(INACT) value of 0.17 min(-1). Spectral scanning of CYP2D6 with paroxetine yielded an increase in absorbance at 456 nm suggesting paroxetine inactivation of CYP2D6 via the formation of a metabolite intermediate complex. This pattern is consistent with the metabolism of the methylenedioxy substituent in paroxetine; such substituents may produce mechanism-based inactivation of cytochrome P450 enzymes. In contrast, quinidine and fluoxetine, both of which are inhibitors of CYP2D6 activity, did not exhibit a preincubation-dependent increase in inhibitory potency. These data are consistent with mechanism-based inhibition of CYP2D6 by paroxetine but not by quinidine or fluoxetine.