Hydraulic actuators benefit robotic systems as they can produce significant force/torque for their size and are robust. However, their dynamic behavior is highly nonlinear, making high-performance closed-loop control a challenging task. With articulated robotic systems, the associated nonlinear multibody dynamics make the control design task even more challenging. Nonlinear model-based (NMB) control methods can be used to address the system nonlinearities. Among NMB control methods, a number of state-of-the-art control performance improvements have been demonstrated for hydraulic manipulators using the virtual decomposition control (VDC) approach. However, all studies on hydraulic systems with VDC have focused on high-inertia and heavy-duty manipulators. In hydraulic cylinder actuated low-inertia and light-weight systems, highly uncertain and hard-to-model nonlinearities, such as actuator friction, can become very dominant in the system’s dynamic behaviour. This paper details the design of a VDC-based controller for a hydraulically actuated light-weight robotic leg. An adaptive friction compensation is incorporated in the control design. The stability of the designed controller is rigorously guaranteed. The experiments with the controller demonstrate a comparable free-space control performance in relation to the state-of-the-art controller for heavy-duty hydraulic manipulators.