The usefulness of various bone mineral measurement techniques is a subject of current controversy. In order to explore whether disparate conclusions may have arisen from differences in analytic methodology, data from published reports of bone mass and nonviolent fractures have been reanalyzed in terms of fracture risk. In the large majority of studies, reduced bone mass was associated with an increased risk of fractures. However, the magnitude of the relationship varied much more among cross-sectional studies than among prospective studies, suggesting that bias related to subject selection and/or postfracture bone loss may have strongly influenced the cross-sectional findings. We conclude that more emphasis should be given to the results of prospective studies, and that more attention should be paid to subject selection in all investigations. Analyzing and presenting results in terms of fracture risk would probably reduce the level of confusion in the field and provide more clinically relevant information. These issues are also applicable to studies of potential fracture risk factors other than bone mass, such as bone structure and bone quality.