||Medical intervention to restore motor function lost due to injury, stroke, or disease is increasingly common. Recent research in this field, known as functional electrical stimulation (FES), has produced a new generation of electrode devices that greatly enhance selectivity of access to neural populations, enabling-for the first time-restoration of motor function approaching what healthy humans enjoy. Research with these devices, however, has been severely hampered by the lack of a stimulation platform and control algorithms capable of exploring their full potential. The following dissertation presents the results of research aimed at addressing this problem. A major theme of this work is the use of software algorithms and analysis principles to facilitate both investigation and control of the motor system. Though many of the algorithms are well known in computer science, their application to the field of motor restoration is novel. Associated with use of these algorithms are important methodological considerations such as speed of execution, convergence, and optimality. The first phase of the research involved development of a hardware and software platform designed to support a wide range of closed-loop response mapping and control routines. Software routines to automate three time-consuming tasks-mapping stimulus thresholds, mapping stimulus-response recruitment curves, and mapping electrode pair excitation overlap- were implemented and validated in a cat model. Computer control, combined with the use of an efficient binary search algorithm, reduced the time need to complete required implant mapping tasks by a factor of 4 or more (compared to manual mapping), making feasible-for the first time-acute experiments investigating multi-array, multijoint experimental limb control. The second phase of the research involved investigating the influence of stimulus timing, within multielectrode trains, on the smoothness of evoked muscle responses. A model for predicting responses was developed and used, in conjunction with function optimization techniques, to identify stimulus timings that minimize response variation (ripple). In-vivo validation demonstrated that low-ripple timings can be identified, and that the influence of timing on ripple depends largely on the response kinetics of the motor unit pools recruited by constituent electrodes. The final phase of the research involved using the response prediction model to simulate the behavior of a feedback-based, stimulus-timing adjustment algorithm. Multiple simulations were executed to assess the influence of three algorithm parameters-filter bandwidth, error sampling delay, and timing adjustment gain-on two performance metrics- convergence time and percent reduction in ripple. Results show that all parameters have an influence on algorithm performance. Convergence speed is the metric most a↵ected by parameter adjustment, improving by a factor of more than 3 (13 cycles to approximately 4 cycles). Ripple reduction is also a↵ected-exhibiting a 17% reduction with appropriate selection of error sampling delay. These results demonstrate the value of using this simulation approach for parameter tuning.