The purpose of our study was to get deeper insights into sprinting with and without running-specific prostheses and to perform a comparison of the two by combining analysis of known motion capture data with mathematical modeling and optimal control problem (OCP) findings. We established rigid multi-body system models with 14 bodies and 16 degrees of freedom in the sagittal plane for one unilateral transtibial amputee and three non-amputee sprinters. The internal joints are powered by torque actuators except for the passive prosthetic ankle joint which is equipped with a linear spring–damper system. For each model, the dynamics of one sprinting trial was reconstructed by solving a multiphase least squares OCP with discontinuities and constraints. We compared the motions of the amputee athlete and the non-amputee reference group by computing characteristic criteria such as the contribution of joint torques, the absolute mechanical work, step frequency and length, among others. By comparing the amputee athlete with the non-amputee athletes, we found reduced activity in the joints of the prosthetic limb, but increased torques and absolute mechanical work in the arms. We also compared the recorded motions to synthesized motions using different optimality criteria and found that the recorded motions are still far from the optimal solutions for both amputee and non-amputee sprinting.