This study evaluated the accuracy of one-element and two-element Hill-type muscle models in predicting the forces of the medial and lateral gastrocnemius muscles in walking and running goats. Goat muscle and tendon data were collected using implanted EMG electrodes, sonomicrometry crystals, and tendon force transducers during various locomotive behaviors. A novel two-element model that accounts for fast and slow motor units provided more accurate predictions of in vivo and in situ muscle forces than a traditional one-element model. However, both models have limitations such as assumptions of independence between activation and force-length/velocity properties.