The main objective of this paper is to highlight the difficulties of identifying shortening and lengthening contractions based on analysis of power produced by resultant joint moments. For that purpose, we present net ankle joint powers and muscle fascicle/muscle-tendon unit (MTU) velocities for medial gastrocnemius (MG) and soleus (SO) muscles during walking in species of different size (humans and cats). For the cat, patterns of ankle joint power and MTU velocity of MG and SO during stance were similar: negative power (ankle moment×angular velocity<0), indicating absorption of mechanical energy, was associated with MTU lengthening, and positive power (generation of mechanical energy) was found during MTU shortening. This was also found for the general fascicle velocity pattern in SO. In contrast, substantial differences between ankle joint power and fascicle velocity patterns were observed for MG muscle. In humans, like cats, the patterns of ankle joint power and MTU velocity of SO and MG were similar. Unlike the cat, there were substantial differences between patterns of fascicle velocity and ankle joint power during stance in both muscles. These results indicate that during walking, only a small fraction of mechanical work of the ankle moment is either generated or absorbed by the muscle fascicles, thus confirming the contribution of in-series elastic structures and/or energy transfer via two-joint muscles. We conclude that ankle joint negative power does not necessarily indicate eccentric action of muscle fibers and that positive power cannot be exclusively attributed to muscle concentric action, especially in humans. © 2013 Elsevier Ltd.