The activity of imipramine 2-hydroxylase highly correlated with that of desipramine 2-hydroxylase but not with that of desipramine N-demethylase. The correlation was also found between N-demethylation and 2-hydroxylation when imipramine was used as a substrate, whereas no correlation was observed between them when desipramine was used in place of imipramine. Both activities of desipramine and imipramine 2-hydroxylase were markedly inhibited by quinidine but not by quinine. Although the activity of imipramine N-demethylase was slightly inhibited by both quinidine and quinine, the activity of desipramine N-demethylase was unaffected under the same conditions. The activity of imipramine N-demethylase was roughly correlated with the amounts of P450 3A4 immunochemically determined and the activities of testosterone 6 beta-hydroxylase in human liver microsomes. The P450 3A4 catalyzed imipramine N-demethylation much more efficiently than 2-hydroxylation in a reconstituted system, whereas neither N-demethylation nor 2-hydroxylation of desipramine was catalyzed by P450 3A4. The activity of imipramine N-demethylase was inhibited, to various extents, by anti-P450 3A4 antibodies in human liver microsomes. Taking together these and other results, it is suggested that P450 3A4, other than P450 2Cmp, also partly contributes to N-demethylation of imipramine, depending on human liver microsomes.