The purpose of this investigation was to determine the cross-sectional geometry of the radius in female and male cadaveric specimens using dual-energy X-ray absorptiometry (DXA), to measure the accuracy of this technique compared with a digitizing procedure, and to measure the correlation between these DXA-based geometric variables and the load required to produce a forearm fracture. Paired intact forearms were scanned at a distal site and at a site approximately 30% of the forearm length from the distal end. The cross-sectional area and the moments of inertia of two sections at 10 and 30% of the forearm length were computed from the X-ray attenuation data. One member of each pair was then sectioned at the 30% location, which is mostly cortical bone, and the section was traced on a digitizing pad. The other forearm was loaded to failure in a servohydraulic materials test system. The DXA-based area and moment of inertia at 30% correlated significantly with the digitized results (r²=0.93 for area;​ r²=0.95 for moment;​ P<0.001). The conventional bone mineral density from DXA did not associate significantly with failure load, but the minimum moment of inertia and the cross-sectional area at 10% correlated in a strong and significant manner with the forearm fracture force (r²=0.67 for area;​ r²=0.66 for moment;​ P<0.001). The determination of radial bone cross-sectional geometry, therefore, should have better discriminatory capabilities than bone mineral density in studies of bone fragility and fracture risk.