The purposes of this study were (1) to define and estimate the direction and amount of the energy transfer between the knee and ankle through gastrocnemius (GA) and plantaris (PL) muscles during cat locomotion, and (2) to test the assumption that the force and activity patterns of soleus (SO), GA, and PL are mechanically and physiologically advantageous for providing the transfer of energy between these joints. The direction, amount and rate of the energy transfer through a two-joint muscle were defined using a theoretical analysis of movements in two adjacent joints spanned by the two-joint muscle. The energy transferred between the ankle and the knee was calculated using the time integration of the difference between the power developed by the moments of SO, GA, and PL at the ankle joint and the total power of these muscles. The total power of SO, GA, and PL muscles, and the power of their movements about the ankle and knee, were obtained using the experimentally determined muscle forces, the rates of change in muscle length, and the angular velocities at the knee and ankle which were calculated from the kinematics and the geometry of the cat hindlimb. Muscular forces and hindlimb kinematics of the cats were recorded during normal walking and trotting on a treadmill at speeds of 0.4, 0.8, 1.2, 1.5, and 1.8 ms-1 using 'E'-shaped tendon transducers and high-speed video, respectively. It was found that during the early phase of support, there was a transfer of mechanical energy from the ankle to the knee through GA and PL. During the late phase of support, mechanical energy was transferred from the knee to the ankle. The amount of energy transferred increased with increasing speeds of locomotion. The energy transferred from the ankle to the knee was 3-60 mJ (7-22% of the negative work done by the moments of SO, GA, and PL at the ankle), and the energy transferred from the knee to the ankle was 10-67 mJ (7-14% of the positive work done by the moments of SO, GA, and PL at the ankle). The results of this study suggest that the activation and the forces of one-joint SO and multi-joint GA and PL are organized in such a way as to fit the features of the design of these ankle extensor muscles in order to provide locomotion efficiently. For example, the decrease in the contractile abilities of SO during the late phase of support at fast speeds of locomotion may be compensated for by the transfer of energy from the knee to the ankle through GA and PL. The design of GA and PL (a high percentage of fast-twitch muscle fibers, large angles of pinnation and short length of the fibers, long tendons, and the location about the ankle and knee joints) seems to be well suited for transferring mechanical energy between the ankle and knee at fast speeds of locomotion. Because of the design of GA and PL, their contractile abilities remain close to the maximum at fast speeds of locomotion. The design of GA and PL allows for extension of the ankle joint through the action of the knee extensor muscles during knee extension with a relatively small change in length of GA and PL.