High resolution, peripheral quantitative computed tomography (HR-pQCT) scanners can now characterize an individual's trabecular architecture, cortical structure, and volumetric bone mineral density at a nominal resolution of 61 μm. While predictions of failure load of the distal radius and tibial diaphysis in compression by finite element analysis (FEA) of HR-pQCT scans have been validated against mechanical tests of cadaveric bones in compression, namely for images with nominal resolutions of 82 μm and 165 μm, the HR-pQCT parameters that best predict bending strength of cortical bone remain unknown. Therefore, we scanned cadaveric forearms from 31 elderly donors (Female:​ 72.8 ± 8.8 years and Male:​ 72.1 ± 6.3 years), and then loaded the radial diaphysis to failure in three-point bending after denuding each bone (38 in total). The cortical parameters had stronger correlations with ultimate moment than the trabecular parameters such that cortical area and estimated failure load of the distal radius had the highest Spearman correlation coefficients (r = 0.89 and r = 0.81, respectively, p < 0.0001). Despite being a known determinant of bone strength, cortical porosity of the distal radius did not correlate with ultimate moment (p = 0.8537). In multivariate linear regressions with section modulus (SM) of the radial diaphysis as one of two predictors of bending strength, cortical area and cortical thickness were each significant contributors to the prediction of ultimate moment. Their contribution was one-half and one-third, respectively, of the contribution from SM. None of the HR-pQCT parameters were strongly correlated with post-yield displacement, an indicator of bone brittleness. In support of HR-pQCT imaging of the distal radius to identify individuals with osteoporosis, the present study found that parameters of the cortex and failure load predictions by linear FEA are strongly related to the bending strength of cortical bone.