In this review, we have made use of some simple engineering concepts to summarize current efforts relating QCT measures to bone density and strength. From a variety of in vitro experiments on cadaveric vertebrae and femora, it is evident that both apparent and ash densities are strong linear functions of QCT measures, with coefficients of determination ranging from 0.49 to 0.90 and relative errors from 44.9% to 7.1%. QCT data also can be used (with somewhat less confidence) to determine the compressive modulus (R2s from 0.36 to 0.68, relative errors from 45.0% to 35.5%) and compressive strength (R2s from 0.58 to 0.70, relative errors from 56.5% to 39.9%) of trabecular bone from the proximal femur and vertebral body. In cortical bone, material properties are only correlated weakly with QCT measures. Experiments designed to relate QCT data to failure loads for the proximal femur and vertebral body have been remarkably successful. Coefficients of determination have ranged from 0.32 to 0.93, with relative errors from 31.1% to 13.9%. However, when the in vitro failure loads determined in these experiments are compared against available estimates of in vivo loads on the spine and hip, it is apparent, at least in the elderly, that in vivo loads are relatively close to those that cause fracture in vitro. To assess the possibility of developing QCT-based clinical predictors of fracture risk for individual patients, we have introduced the concept factor of risk often used in engineering design to account for uncertainties in estimates of service loads and component strength. The factor of risk for a particular loading condition is defined as the ratio of expected service loads to the known failure loads. To extend this concept to densitometric fracture risk prediction in vivo, it is important to recognize that densitometric data must not only be used to predict the ultimate load carrying capacity of the region of interest, but that this ultimate load must then be compared to the forces expected in vivo under comparable loading conditions. One difficulty with this approach is that little is known about the in vivo forces that are associated with atraumatic age-related fractures of the hip and vertebrae and even less about the forces applied to the hip and spine during traumatic events such as falls. However, from available estimates of in vivo loads during bending and lifting, it is apparent that in the elderly, factors of risk for the spine can easily approach 1.(ABSTRACT TRUNCATED AT 400 WORDS)