OBJECTIVE Tamoxifen is an effective treatment for metastatic and primary breast cancer and is now being evaluated as a chemoprevention agent in healthy women. Any long-term effects on estrogen-sensitive tissues such as bone may have important therapeutic implications. METHODS We measured bone mineral density (BMD) in the lumbar spine and hip using dual-energy x-ray absorptiometry (DXA) in premenopausal and postmenopausal healthy women who participated in our placebo-controlled tamoxifen chemoprevention of breast cancer trial. RESULTS BMD data are now available from 179 women for this analysis. In premenopausal women, BMD decreased progressively in the lumbar spine (P < .001) and in the hip (P < .05) for women on tamoxifen, but not those on placebo. The mean annual loss in lumbar BMD per year over the 3-year study period in tamoxifen-treated compliant women who remained premenopausal throughout the study period was 1.44% (1.88% calculated on an intent-to-treat basis) compared with a small gain of 0.24% per annum for women on placebo (P < .001). Tamoxifen had the opposite effect in postmenopausal women. The mean annual increase in BMD for women on tamoxifen was 1.17% in the spine (P < .005) and 1.71% in the hip (P < .001) compared with a noninsignificant loss for women on placebo. CONCLUSIONS These results indicate that tamoxifen treatment is associated with a significant loss of BMD in premenopausal women, whereas it prevents bone loss in postmenopausal women. These adverse and beneficial effects of tamoxifen should be considered in the assessment of the therapeutic benefits for both the adjuvant treatment and the chemoprevention of breast cancer.