Fractures of the proximal femur represent a national health problem of crisis proportions. More than 250,000 hip fractures occur annually, resulting in costs in excess of 7 billion dollars per year. While there is growing evidence that certain therapeutic measures such as estrogen replacement can help retard bone loss and thus reduce hip fracture incidence, some treatment modalities are themselves associated with significant risks. It becomes increasingly important, therefore, to identify only those patients at greatest risk of fracture so that appropriate therapy can be instituted. New diagnostic imaging modalities such as quantitative computed tomography (QCT) could be employed for this purpose if appropriate tissue characterization procedures and fracture risk predictors are developed. Hence, the objective of this research was to investigate the mechanisms of proximal femur fracture through the use of QCT coupled with the finite element method of structural analysis, from which verified biomechanical predictors of hip fracture risk could be developed. To address this goal, it was initially imperative to study the material properties of both the trabecular and cortical bone within the proximal femur and to develop OCT based characterization techniques. Thus, four pairs of fresh human proximal femora were imaged with a conventional CT scanner from which fifty cylindrical trabecular bone specimens were fabricated and tested in compression. It was demonstrated that bone density (R2-0.73), elastic modulus (R2=0.89) and compressive strength (R2=0.89) could be determined accurately and noninvasively by using single-energy QCT. In addition, the material properties of the metaphyseal shell were investigated through the use of three-point bending tests with 150 prismatic plate specimens. Both the elastic modulus and tensile strength were shown to be reduced from that measured within the femoral diaphysis by 33 and 21 percent respectively. Though the majority of this reduction could be attributed to density, architectural differences were also implied by the observed decrease in the ratio of longitudinal to transverse moduli (2.1 diaphyseal, 1.8 metaphyseal) and strength (2.7 diaphyseal, 2.0 metaphyseal). Structural analysis was subsequently performed on two intact femora using the displacement based finite element method. Three dimensional, linear and nonlinear finite element models were noninvasively generated for the intact bones by using geometry and material property data obtained from QCT images and the appropriate OCT-material property regressions derived previously. Each bone was then instrumented with strain gages and tested to failure in one of either two loading configurations; a simplified one-legged stance or fall. For both loading conditions, the maximum calculated von Mises effective strain from the linear analysis was predictive of in-vitro bone failure to within 8 percent. The results of the nonlinear analyses also exhibited excellent agreement with the in-vitro fracture location and failure load, but in addition provided insight into the fracture process by demonstrating internal trabecular failure and subsequent load transferal to the metaphyseal shell just prior to structural collapse. During one-legged stance, the critical failure strains were developed within the primary tensile trabeculae at the subcapital region of the femoral neck. In contrast, for the simulated fall, critical failure strains were observed within the intertrochanteric region, suggesting that QCT measurements at this sight would be a sensitive predictor of intact bone strength during falls. To verify this hypothesis, 12 intact human femora were scanned by OCT and subsequently tested to failure under the simulated fall conditions. A highly significant positive linear correlation was observed between the load at bone failure and the product of the average trabecular bone OCT value and total bone cross sectional area as measured from a OCT image made through the intertrochanteric region (R420.93). Similarly good correlation was observed when using average intertrochanteric trabecular QCT data alone (R2.0.89). In conclusion, these results demonstrate that intact bone failure can be accurately modeled using the finite element method and that for the particular fall loading configuration studied, intact bone strength can be predicted noninvasively with high accuracy using QCT.