Large cyclic movements between the femoral stem and bone during the first weeks after total hip arthroplasty may hamper bone ingrowth and adversely affect the eventual success of the arthroplasty. Little is known, however, about the magnitude of the motions and its relationship to design and surgical factors. A two-dimensional finite element model of a cementless prosthesis inserted into the proximal femur was constructed to study the effects of two mechanical variables--the stiffness of the implant and the coefficient of friction between bone and implant--on the magnitude of the motions. We investigated the influences of these variables on the subsidence of the prosthesis, the magnitudes of the cyclic motions, and the level of the interface stresses. The presence of friction reduced cyclic motions by about 85% compared with a frictionless interface. Once friction was assumed, varying the coefficient of friction had little effect. The effect of friction on the interface stress state and gross subsidence of the prosthesis was not as great as on cyclic motion. Implant stiffness also affected the magnitudes and distributions of the cyclic motions along the interface. A flexible stem generated motions about three to four times larger proximally than those of a stiff stem, which generated larger motions distally. The influence of stem stiffness on interface stresses and prosthetic subsidence was less than on cyclic motion. The location of the peak shear stresses at the interface around a bonded prosthesis corresponded to the location where cyclic interface motion was maximal for an unbonded prosthesis. However, no direct relationship was found between the magnitudes of peak stresses and the amplitudes of cyclic motions.