Roxithromycin has been shown to be a relatively weak inhibitor of cytochrome P450 (P450 or CYP)-dependent drug oxidations, compared with troleandomycin. The potential for roxithromycin and its major metabolites found in human urine [namely the decladinosyl derivative (M1), O-dealkyl derivative (M2), and N-demethyl derivative (M3)] to inhibit testosterone 6beta-hydroxylation after metabolic activation by CYP3A4 was examined and compared with inhibition by troleandomycin and erythromycin in vitro. Of roxithromycin and its studied metabolites, M3 was the most potent in inhibiting CYP3A4-dependent testosterone 6beta-hydroxylation by human liver microsomes and was activated to the inhibitory P450.Fe2+-metabolite complex to the greatest extent. Roxithromycin and its metabolites were N-demethylated by human liver microsomes, although the rates were slower than those measured with troleandomycin and erythromycin as substrates. Recombinant human CYP3A4 in a baculovirus system coexpressing NADPH-P450 reductase was very active in catalyzing the N-demethylation of roxithromycin, M1, and M2, as well as troleandomycin, erythromycin, and M3. The order for inhibition of CYP3A4-dependent testosterone 6beta-hydroxylation activities by these macrolide antibiotics in the recombinant CYP3A4 system was estimated to be troleandomycin > erythromycin >/= M3 >/= M2 > M1 >/= roxithromycin. Erythromycin, roxithromycin, and its metabolites all failed to inhibit CYP1A2-dependent (R)-warfarin 7-hydroxylation and CYP2C9-dependent (S)-warfarin 7-hydroxylation but did inhibit CYP3A4-dependent (R)-warfarin 7-hydroxylation. These results suggest that roxithromycin itself is not as potent an inhibitor of CYP3A4 activities as are troleandomycin and erythromycin, probably because of the slower metabolism of this compound to metabolites M1, M2, and M3 in humans.