To improve the treatment of breast cancer, there has been a need for alternative aromatase inhibitors (AIs) that bring about adequate aromatase inhibition, while limiting side effects. Since two tamoxifen metabolites have been documented as AIs, we tested a wide range of tamoxifen metabolites on aromatase in order to better understand structural interactions with aromatase and constructed structure-function relationships as a first step toward the development of novel inhibitors. The ability of ten tamoxifen metabolites to inhibit recombinant aromatase (CYP19) was tested using microsomal incubations. The selectivity of the most potent aromatase inhibitor identified, norendoxifen, was characterized by studying its ability to inhibit CYP450 enzymes important in clinical drug-drug interactions, including CYP2B6, 2C9, 2C19, 2D6, and 3A. Computerized molecular docking with the X-ray crystallographic structure of aromatase was used to describe the detailed biochemical interactions involved. The inhibitory potency order of the tested compounds was as follows: norendoxifen ≫ 4,4'-dihydroxy-tamoxifen > endoxifen > N-desmethyl-tamoxifen, N-desmethyl-4'-hydroxy-tamoxifen, tamoxifen-N-oxide, 4'-hydroxy-tamoxifen, N-desmethyl-droloxifene > 4-hydroxy-tamoxifen, tamoxifen. Norendoxifen inhibited recombinant aromatase via a competitive mechanism with a K ( i ) of 35 nM. Norendoxifen inhibited placental aromatase with an IC(50) of 90 nM, while it inhibited human liver CYP2C9 and CYP3A with IC(50) values of 990 and 908 nM, respectively. Inhibition of human liver CYP2C19 by norendoxifen appeared even weaker. No substantial inhibition of CYP2B6 and CYP2D6 by norendoxifen was observed. These data suggest that multiple metabolites of tamoxifen may contribute to its action in the treatment of breast cancer via aromatase inhibition. Most of all, norendoxifen may be able to serve as a potent and selective lead compound in the development of improved therapeutic agents. The range of structures tested in this study and their pharmacologic potencies provide a reasonable pharmacophore upon which to build novel AIs.