It is not clear which non-invasive method is most effective for predicting strength of the proximal femur in those at highest risk of fracture. The primary aim of this study was to compare the abilities of dual energy X-ray absorptiometry (DXA)-derived aBMD, quantitative computed tomography (QCT)-derived density and volume measures, and finite element analysis (FEA)-estimated strength to predict femoral failure load. We also evaluated the contribution of cortical and trabecular bone measurements to proximal femur strength. We obtained 76 human cadaveric proximal femurs (50 women and 26 men;​ age 74 ± 8.8 years), performed imaging with DXA and QCT, and mechanically tested the femurs to failure in a sideways fall configuration at a high loading rate. Linear regression analysis was used to construct the predictive model between imaging outcomes and experimentally-measured femoral strength for each method. To compare the performance of each method we used 3-fold cross validation repeated 10 times. The bone strength estimated by QCT-based FEA predicted femoral failure load (R²_adj = 0.78, 95%CI 0.76–0.80;​ RMSE = 896 N, 95%CI 830–961) significantly better than femoral neck aBMD by DXA (R²_adj = 0.69, 95%CI 0.66–0.72;​ RMSE = 1011 N, 95%CI 952–1069) and the QCT-based model (R²_adj = 0.73, 95%CI 0.71–0.75;​ RMSE = 932 N, 95%CI 879–985). Both cortical and trabecular bone contribute to femoral strength, the contribution of cortical bone being higher in femurs with lower trabecular bone density. These findings have implications for optimizing clinical approaches to assess hip fracture risk. In addition, our findings provide new insights that will assist in interpretation of the effects of osteoporosis treatments that preferentially impact cortical versus trabecular bone.