Metabolic energy expenditure (MEE) is commonly used to characterize human motion. In this study, a general joint-space dynamic model of MEE is developed by integrating the principles of thermodynamics and multibody system dynamics in a joint-space model that enables the evaluation of MEE without the limitations inherent in experimental measurements or muscle-space models. Muscle-space energetic components are mapped to the joint space, in which the MEE model is formulated. A constrained optimization algorithm is used to estimate the model parameters from experimental walking data. The joint-space parameters estimated directly from active subjects provide reliable estimates of the trend of the cost of transport at different walking speeds. The quantities predicted by this model, such as cost of transport, can be used as strong complements to experimental methods to increase the reliability of results and yield unique insights for various applications.