Prospective studies have shown that bone mass, measured by any method and at any site, is inversely related to a woman's risk of fracture. Most types of fractures are more frequent in women with low bone mass. Bone mineral density measured in the proximal femur appears to have a stronger relationship to risk of hip fracture than bone density measured in other sites. On the other hand, measurement of bone mass in the spine does not appear to be substantially superior to measurements of bone mass in other sites. Even after age 80, bone mass measurements have a strong predictive value for most types of fracture. A woman's lifetime risk of hip fracture can be estimated from bone mass measurements made in the perimenopausal period. Although an individual's level of bone mass may vary from site to site, the incremental value of measuring bone mass in more than one site is not certain. Bone mass measurements, perhaps in combination with other risk factors for fracture, help define individuals with a high risk of fracture who stand to benefit most from efforts to prevent fractures.