The normal human running has been simulated by two-dimensional biped model with 7 segments. Series of normal running experiments were performed and data of ground reaction forces measured by force plate was analyzed and was fitted to some Fourier series. The model is capable to simulate running for different ages and weights at different running speeds. A proportional derivative control algorithm was employed to grant stabilization during each running step. For calculation of control algorithm coefficients, an optimization method was used which minimized cinematic differences between normal running model and that of the experimentally obtained from running cycle data. This yielded the estimated torque coefficients of the different joints. The estimated torques and the torque coefficients were then applied to specific below-knee prosthesis (a SACH foot) to simulate healthy-running motion of joints. Presently the SACH foot is designed for amputee’s walking; our data was used to modify such construct for running purposes. The goal was to minimize the differences between normal human model and a subject wearing a SACH foot during running. Kinematical curves of models for the obtained optimum mechanical properties indicated that prosthetic leg can reasonably produce the kinematics of normal running under normal joint driving torques.