This work evaluated the kinetic behavior of fluoxetine O-dealkylation in human liver microsomes from different CYP2C19 genotypes and identified the isoenzymes of cytochrome P450 involved in this metabolic pathway. The kinetics of the -trifluoromethylphenol (TFMP) formation from fluoxetine was determined in human liver microsomes from three homozygous (wt/wt) and three heterozygous (wt/m1) extensive metabolizers (EMs) and three poor metabolizers (PMs) with m1 mutation (m1/m1) with respect to CYP2C19. The formation rate of TFMP was determined by gas chromatograph with electron-capture detection. The kinetics of TFMP formation was best described by the two-enzyme and single-enzyme Michaelis-Menten equation for liver microsomes from CYP2C19 EMs and PMs, respectively. The mean intrinsic clearance (Vmax/Km) for the high- and low-affinity component was 25.2 l/min/nmol and 3.8 l/min/nmol of cytochrome P450 in the homozygous EMs microsomes and 12.8 l/min/nmol and 2.9 l/min/nmol of cytochromecytochrome P450 in the heterozygous EMs microsomes, respectively. Omeprazole (a CYP2C19 substrate) at a high concentration and triacetyloleandomycin (a selective inhibitor of CYP3A4) substantially inhibited O-dealkylation of fluoxetine. Furthermore, fluoxetine O-dealkylation was correlated significantly with S-mephenytoin 4-hydroxylation at a low substrate concentration and midazolam 1 -hydroxylation at a high substrate concentration in liver microsomes of 11 Chinese individuals, respectively. Moreover, there were obvious differences in the O-dealkylation of fluoxetine in liver microsomes from different CYP2C19 genotypes and in microsomal fractions of different human-expressed lymphoblast P450s. The results demonstrated that polymorphic CYP2C19 and CYP3A4 enzymes were the major cytochrome P450 isoforms responsible for fluoxetine O-dealkylation, whereas CYP2C19 catalyzed the high-affinity O-dealkylation of fluoxetine, and its contribution to this metabolic reaction was gene dose-dependent.