Nerve root stimulators for spinal cord injured patients were first developed in the 1970s in London and are now well established with the Sacral Anterior Root Stimulator Implant (SARSI) for bladder and bowel control. Experience with lumbar root stimulators (LARS I) led to a rethink of both the purpose of the device (through an improved understanding of patients' expectations) and of the stimulator's electronic design. Paraplegics willing to receive a SARSI would also benefit from leg exercise. Stimulating both sacral and lumbar roots with a single implant would combine both applications at little extra cost as placement would only require one operation. Developing a new implant offered, via this thesis, an opportunity to advance the knowledge and refine the stimulation method. The primary hypothesis was that independent stimulation channels would improve the selectivity by eliminating cross-talk. This thesis demonstrates and explains the occurrence of cross-talk (with patient's data and animal recordings) and its reduction with the new stimulation method. The design of both the experimental stimulator and the integrated-circuit developed as output stage of the new implant is detailed, complete with circuits schematics. Further, an analysis of the literature on nerve root compression supports an explanation of the mechanisms behind entrapment and compression damages. A study of electrode book impedance follows. It provides an insight into the respective influence of the electrode and mount dimensions and leads to a simple formula to estimate the overall impedance for the IC design. While this work's contribution, as highlighted above, is mainly in the medical engineering domain, a short discussion setting it in a wider context has been added to insist on the human dimension of such a research project.