Centralized controlling approaches, such as full-body trajectory planning and tracking, are prevalent in locomotion control of humanoid robots. How- ever, they cannot be applied to powered prosthetic devices as a dynamic model is not available for the human in the loop, and it is difficult to estimate state of the human. One alternative is based on replay of motion patterns captured from able-bodied human; however, their performance is unknown under large disturbances. In contrast, heuristics-based controllers encode motion patterns in feedback control laws and/or state machines. One such approach called neuromuscular control stems from studies on animal and human locomotion behavior, which has potential to enable diverse and robust locomotion behav- iors in both humanoids and prosthetic devices. As a part of ongoing effort to evaluate this approach on hardware, this thesis presents work on transferring a neuromuscular stance-leg controller to Robotic Neuromuscular Leg 3 (RNL3), a powered segmented leg hardware platform. The controller has been shown in simulation of an idealized 2-segment leg to produce a stable bouncing gait. It is implemented and tested in a physical simulation model of RNL3, then trans- ferred and tested on actual hardware. The controller is capable of maintaining a stable in-place bouncing gait in simulation. However, the robotic leg suffered from mechanical failures during experiments; as a result the controller remains to be verified on hardware.