The objective of this study was to establish the role of loads and prosthesis material properties on the mechanics of the proximal femur after total hip arthroplasty. We developed a three‐dimensional finite element model of an intact human femur and the same femur with a conventional collared straight‐stem femoral component. Using published data, we defined two sets of loading conditions:​ one that represented three phases of gait, and one that represented four different extreme loads. The four extreme loads were based on the peak joint contact forces that occur during stair ascent and isometric contraction of various muscle groups. The model was analyzed with three different material properties for the prosthesis, including cobalt‐chromium alloy, titanium alloy, and a carbon fiber‐reinforced polymer (CFRP) laminate. We assumed that the implant was stable, with rigid bonding, collar contact, and no cement. To address femoral component loosening, we examined the shear stresses at the implant‐bone interface;​ to address adaptive bone remodeling, we examined the principal stresses in the supporting cortical bone relative to those in the intact femur. Our analyses of the various loading conditions demonstrated large out‐of‐plane bending moments and torsional moments, especially for the load representing stair ascent. Based on stepwise multiple regressions, the maximum shear stresses at the implant‐bone interface in the distal region were dependent on the total applied axial force and torsion;​ the maximum shear stresses in the proximal region were dependent on the axial component of the joint contact force alone. Reduction in the prosthesis stiffness, by substitution of the CFRP material properties, resulted in lower interface shear stresses at the distal end of the stem and higher interface shear stresses at the more proximal sections, consistent with the findings of others. We fit equations, based on composite beam theory, to the maximum implant‐bone interface shear stresses and the cortical bone principal stresses as a function of the axial modulus of the prosthesis. These equations can be used to estimate the maximum stresses at the interface and in the cortical bone that would be predicted by similar models, for the same prosthesis constructed of alternative materials, relative to the stresses in the intact femur. The nonlinear nature of these relationships was such that the cortical bone stresses changed more rapidly, as a function of the prosthesis modulus, for lower values of elastic modulus, especially in the more proximal sections.