OBJECTIVE To determine the anthropometric, historical, and lifestyle factors associated with bone mineral density (BMD) of the spine and proximal femur in older women. METHODS Cross-sectional analyses. METHODS Four clinical centers in Baltimore, Maryland; Minneapolis, Minnesota; Portland, Oregon; and the Monongahela Valley, Pennsylvania. METHODS 7963 ambulatory, nonblack women 65 year of age or older. METHODS Medical history was obtained by questionnaire and interview, and physical and anthropometric data were obtained by examination. Lumbar spine and proximal femoral BMDs were measured using dual-energy x-ray absorptiometry. RESULTS The multivariable models could predict 21% and 25% of the difference between participants in BMD at the femoral neck and lumbar spine, respectively. Weight was most highly associated with BMD. Postmenopausal estrogen use and other indicators of total estrogen exposure were strongly associated with increased BMD. Use of diuretics (both thiazide and nonthiazide), activity levels and muscle strength, alcohol intake, and dietary calcium intake were associated with higher BMD. A family history of osteoporotic fracture was strongly associated with low BMD. European ancestry and blond hair, childbirth or breast feeding, a history of hyperthyroidism, and progestin use were not associated with axial BMD. CONCLUSIONS Weight is strongly associated with BMD. Estrogen exposure, physical activity, and calcium intake are also positively associated with BMD, whereas a family history of osteoporosis is associated with reduced BMD. These associations suggest ways to better identify risk for fracture.