Hip fracture is a serious and common injury that can lead to permanent disability, pneumonia, pulmonary embolism, and death. Research to help prevent these fractures is essential. Computed tomographic (CT) scan-based finite element (FE) modeling is a tool that can predict proximal femoral fracture loads in vitro. Because this tool might be used in vivo, this study examined whether FE models generated from CT scans in situ and in vitro yield comparable predictions of proximal femoral fracture load. CT scans of the left proximal femur of two human cadavers were obtained in situ and in vitro, and three-dimensional FE models employing nonlinear mechanical properties were generated from each CT scan. The models were evaluated under single-limb stance-type loading by applying displacements incrementally to the femoral head. The FE-predicted fracture load (F_FE) was the maximum femoral head reaction force. F_FE for the in situ-derived models for the two subjects were 5.2 and 13.3% greater than for the in vitro-derived models. These results demonstrate that using CT scan data obtained in situ instead of in vitro to generate FE models can lead to substantially different predicted fracture loads. This effect must be considered when using this technology in vivo.