The inversely-nonlinear relationship of muscle contraction force and the possible contraction duration (i.e., endurance) is utilized in a method to mathematically predict individual muscle forces. The method uses a nonlinear optimization scheme to solve the redundant distribution problem at the joints of interest. The method is demonstrated at the elbow during isometric contraction and in the lower extremity during locomotion. During gait, the observed muscle activity pattern in the lower extremity (as determined by EMG) shows substantial agreement with that activity pattern predicted when endurance is used as the optimization criterion. The importance of selecting muscle prediction criteria based on sound physiological bases rather than on an arbitrary or mathematically convenient basis is emphasized.