Femoral fractures are a major health issue. Most experimental and finite element (FE) fracture studies use polymethylmethacrylate cups on the greater trochanter (GT) to simulate fall impact loads. However, in vivo fall studies showed that the femur is loaded distally from the GT. Our objective was to incorporate in vivo fall data in FE models to determine the effects of loading position and direction, and size of simulated impact site on the fracture load and fracture type for a healthy and an osteoporotic femur. Twelve sets of loading position and angles were applied through 'near point loads' on the models. Additional simulations were performed with 'cup loads' on the GT, similar to the literature. The results showed no significant difference between fracture loads from simulations with near point loads distally from the GT and those with cup loads on the GT. However, simulated fracture types differed, as near point loads distally from the GT generally resulted in various neck fractures, whilst cup load simulations predicted superior neck and trochanteric fractures only. This study showed that incorporating in vivo fall assessments in FE models by loading the models distally from the GT results in prediction of realistic fracture loads and fracture types.