The female sex hormone estradiol (1) has a variety of beneficial and detrimental effects in women. The triphenylethylene class of non-steroidal estrogens (e.g., tamoxifen, 2) shows tissue-dependent expression of estrogen agonist and antagonist activity and may represent a significant advance over conventional hormone replacement therapy with 1 for prevention of osteoporosis and cardiovascular disease in postmenopausal women. The estrogen receptor is a ligand-activated transcription factor that belongs to the steroid/retinoid family of DNA-binding intracellular receptors (ICR). Studies with deletion and point-mutated receptors have revealed two independent transcription activation domains (AF-1 and AF-2) within the receptor that allow the expression of cell- and promoter-specific agonist activity in transient cotransfection experiments in vitro. We formulated the hypothesis that the tissue-selective profile of 2 was due to induction of a unique receptor conformation in which the antagonist activity in some tissues was due to disruption of AF-2, mediated by a H-bond interaction with the receptor protein in the region of the putative AF-2 α-helix, and the agonist activity in other tissues was a result of a functional AF-1 domain. Combining this hypothesis with analysis of the in vitro and in vivo pharmacology of non-steroidal estrogens, it was proposed that the stilbene portion of 2 was required for AF-1 activity leading to agonist activity in bone, and the ethanolamine side chain was responsible for blocking AF-2 activity leading to antagonism in the uterus. Using this hypothesis, we synthesized analogs of 2 in which the ethanolamine side chain was replaced by alternate H-bond acceptor groups and the degree of conformational freedom was reduced. Among these analogs, acid 6 showed significantly lower maximal residual agonist activity (E_max) in human uterine Ishikawa cells compared to 2. In vivo, 6 tested as an estrogen antagonist in the rat uterus with significantly less residual agonist activity than 2. In ovariectomized rats, 6 demonstrated full agonist activity by maintaining bone mineral density (BMD) at the levels of sham-operated rats through inhibiting bone resorption. The identification of 6 as an analog of 2 with reduced uterine agonist activity that maintained its full agonist activity in bone adds credence to our hypothesis that triphenylethylenes code for expression of AF-1 and AF-2 activity through separate parts of the molecule. Given that most members of the steroid/retinoid receptor family possess multiple transcription activation domains and structural homology with the estrogen receptor, these design principles hold promise for developing tissue- and promoter-selective ligands for other receptors.