Bone graft subsidence is a serious complication of interbody spinal fusion. In this study, 66 mechanical tests were performed on 35 thoracic vertebral bodies to investigate the in situ mechanics of interbody spinal fusion. The relationships among trabecular bone density, bone strength, and size of bone graft area were analyzed. All vertebral bodies were scanned by quantitative computer tomography (QCT) to determine their bone density before mechanical testing. The decorticated trabecular beds of the vertebral bodies, void of all posterior elements, were loaded in a manner similar to that which occurs after surgical interbody fusion. That is, rectangular blocks of polymethylmethacrylate, representing bone grafts, were used to transfer controlled compressive loads to the decorticated vertebral trabecular surface. Both destructive and nondestructive tests were performed. The relationship between QCT bone density and trabecular bone strength was related by a power function, and, on average, the bone density and trabecular bone strength were 0.137 g/cm3 and 3.97 MPa, respectively. Eighty percent of the vertebral bodies with graft covering 25% of the total end plate area or less falled at loads less than 600 N, while 88% of the vertebral bodies with 30% or greater covered were able to carry a load greater than 600 N. The results suggest that the intrinsic behavior of trabecular bone loaded within the vertebral body is little different from the behavior of the whole vertebral body, that QCT bone density is indicative of bone strength, and that interbody graft area should be significantly greater than 30% of the total end plate area to provide a margin of safety. Additionally, the relationship between QCT bone density and bone strength permits generation of a family of curves to predict critical minimum graft area based on QCT measurements and anticipated physiological loads or scaled to patient weight.