I n 29 women with postmenopausal osteoporosis, the proportion of total bone surface undergoing resorption or formation was evaluated by microradiography of iliac crest biopsy samples before and after short-term (2½-4 months) and long-term (26-42 months for estrogen and 9-15 months for anabolic hormone) treatment. After estrogen administration, values for bone-resorbing surfaces decreased, although less prominently after long-term than after short-term therapy. The magnitude of this decrease was positively correlated with the pretreatment value for bone-resorbing surfaces (P < 0.001). When the pretreatment value for bone-resorbing surfaces was used as a covariable, estrogen and anabolic hormone appeared to be equally effective. For bone-forming surfaces, short-term therapy with either hormone had no effect but long-term therapy significantly decreased the values. Serum immunoreactive parathyroid hormone (IPTH) increased significantly after estrogen therapy; the change in IPTH was inversely related to the change in serum calcium (P < 0.001, sign test). We conclude that the primary effect of sex hormones in postmenopausal osteoporosis is to decrease the increased level of bone resorption, perhaps by decreasing the responsiveness of bone to endogenous parathyroid hormone. However, this favorable effect, at least in part, is negated after long-term treatment by a secondary decrease in bone formation. Our data are consistent with the concept that the maximal benefit that can be derived from sex hormone therapy in postmenopausal osteoporosis is arrest or slowing of the progession of bone loss.