Techniques for measuring bone mineral content (BMC) were developed for the purpose of providing an objective and noninvasive indication of bone strength (or lack thereof) and fracture risk, to the extent that strength relates to bone mass. As such, BMC measurements could help to (1) identify those who are most likely to experience nonviolent fractures in the future and who would therefore benefit most from preventive measures, (2) improve their treatment compliance, and (3) monitor the efficacy of treatments intended to reduce bone loss. All these potential uses require that the measurement provide an indication of fracture risk (probability of fractures). During the past 10-15 years there have been conflicting reports regarding the association of reduced BMC with nonviolent fractures. Some authors have criticized the usefulness of BMC measurements, whereas others have questioned the value of one or more techniques. However, the epidemiology of osteoporosis has only recently been subjected to rigorous study. The use of appropriate statistical methods for relating fracture risk to bone mass may be no more widely practiced in osteoporosis epidemiology today than it was for studying risk factors (e.g., blood pressure) in cardiovascular epidemiology during the 1960s. The intent of this article is to explore three areas that may have contributed to controversy in the study of bone mass and fracture occurrence: (1) perspective of the investigators, (2) study design, and (3) analytic methodology. Although the focus of this paper is on bone mass, these considerations are equally applicable to some investigations of other risk factors for osteoporotic fractures (e.g., bone architecture, bone turnover and loss rate, or biochemical markers of bone loss).