(R)- and (S)-fluoxetine were found to be competitive inhibitors of P450 2D6-mediated bufuralol 1'-hydroxylation in vitro, yielding Ki values of 1.38 +/- 0.48 and 0.22 +/- 0.11 microM, respectively. Their N-demethylated metabolites were also found to be potent inhibitors (Ki, (R)-norfluoxetine, 1.48 +/- 0.27 microM; (S)-norfluoxetine, 0.31 +/- 0.04 microM). The microsomal (R)- and (S)-fluoxetine N-demethylase activities for 14 human liver samples were on average 29.6 +/- 13.5 and 19.4 +/- 11.8 pmol of product/min/mg of protein, respectively. The individual rates of N-demethylation correlated with microsomal immunodetectable P450 2D6 levels; (R)-fluoxetine, r = 0.64, P < .05; (S)-fluoxetine, r = 0.63, P < .05. However, this correlation was significantly weaker than the excellent correlation obtained for P450 2D6-marker bufuralol 1'-hydroxylase activity and P450 2D6 levels (r = 0.92, P < or = .01). Quinidine, a potent inhibitor of P450 2D6, inhibited the demethylation of each enantiomer by only approximately 20% at a concentration 300 times greater than the Ki determined for the quinidine inhibition of bufuralol 1'-hydroxylase. Furthermore, antiserum recognizing P450 2D6 inhibited 82% of microsomal bufuralol 1'-hydroxylase activity but only 27% of the (R)-fluoxetine N-demethylase activity in the same human liver sample. In summary, these data indicate that the enantiomers of fluoxetine and norfluoxetine are potent inhibitors of P450 2D6 and that P450 forms other than P450 2D6 appear to be responsible for the majority of microsomal fluoxetine N-demethylation.