The limitations of empirical models for human metabolic energy expenditure (MEE) have motivated the search for general-purpose models based on investigations of heat dissipation. As opposed to the muscle-space approach to MEE modeling, the study outlined here adopts a joint-space MEE model developed from thermodynamics and multibody system dynamics principles. The MEE is formulated as a function of joint-space kinematic and kinetic variables and commonly measured system kinetic and kinematic parameters. This study builds on previous work by introducing an updated heat dissipation formulation that contains additional kinetic variables and accounts for the different energetic costs of negative and positive work. A multiple regression analysis is used to develop a 12 parameter MEE model with experimental MEE data obtained from human steady-state walking experiments. The resulting model was able to reliably predict the MEE of different subjects walking over a large range of speeds.