Neuromodulation of spinal sensorimotor circuits improves motor control in animal models and humans with Spinal Cord Injury (SCI) and Parkinson disease. Stimulation parameters are tuned manually and remain constant during motor execution which is suboptimal to mediate maximum therapeutic effects. Here, I present a novel neuroprosthetic system that enabled adaptive changes of neuromodulation parameters during locomotion and allowed to restore high-fidelity control over leg movements in paralyzed rats. Beyond the therapeutic potential, these findings provide a conceptual and technical framework to personalize neuromodulation treatments for other neurological disorders. Several limitations have restricted the development of neuroprosthetic systems for closed loop neuromodulation. (1) First, it required a mechanistic understanding of the relationships between stimulation features and the recruitment of specific sensorimotor circuits. I found that electrical neuromodulation primarily recruits afferent reflex pathways that lead to coordinated activity of leg muscles during stepping. Moreover, the specific electrode location on the spinal cord could activate distinct reflex pathways and activate specific leg muscle groups of paralyzed rats. These results have been leveraged for the design of flexible and stretchable multi-electrode arrays for electrical and chemical spinal cord stimulation. (2) Second, it was necessary to perform comprehensive mapping experiments to characterize the effect of neuromodulation parameters on hind limb kinematics in order to establish stable and robust feedback signals for real time control. Step height and ground reaction forces emerged as the primary targets for the control of closed loop neuromodulation after spinal cord injury. (3) Third, implementation and optimization of closed-loop neuromodulation strategies necessitated the development of an advanced technological platform that combined feedback and feed-forward loops that match the natural flow of information in the modulated neural systems. These integrated developments allowed animals with complete spinal cord injury to perform over 1000 successive steps without failure, and to climb staircases of various heights and lengths with precision and fluidity. Moreover, the neuroprosthetic system was able to alleviate locomotor deficits in an alpha-synuclein rodent model of Parkinson’s disease. Current knowledge of human spinal cord properties in response to electrical neuromodulation suggests that the developed control policies can translate into clinical applications to improve neurorehabilitation therapies. Moreover, the developed neuroprosthetic system can readily be interfaced with control signals from the brain to establish cortico-spinal neuroprostheses that are intended to promote activity-dependent plasticity during recovery from spinal cord injury.