Polyimide regenerative electrodes (RE) constitute a promising neural interface to selectively stimulate regenerating fibers in injured nerves. The characteristics of the regeneration through an implanted RE, however, are only beginning to be established. It was recently shown that the number of myelinated fibers distal to the implant reached control values 7 months postimplant; however, the functional recovery remained substantially below normal [J Biomed Mater Res 60 (2002) 517]. In this study we sought to determine the magnitude, and possible selectivity, of axonal regeneration through the RE by counting sensory and motor neurons that were retrogradely labeled from double tracer deposits in the sciatic nerve. Adult rats had their right sciatic nerves transected, and the stumps were placed in silicone tubes; some simply were filled with saline (Tube group), and others held a RE in its center (RE group). Simultaneously, the proximal stump was exposed to Diamidino Yellow. Two months later the nerves were bilaterally excised distal to the implant, and exposed to Fast Blue. Electrophysiological recordings, and skin nociceptive responses confirmed previous findings of partial functional recovery. In controls, an average of 20,000 and 3080 neurons were labeled in L4-L5 dorsal root ganglia (with minor contributions from L3 and/or L6), and in the ventral horn of the lumbar spinal cord, respectively. In the regenerating side, 35% of the DRG neurons were double-labeled, without differences between groups. In contrast, only 7.5% of motoneurons were double-labeled in the RE group, vs. 21% in the Tube group. Moreover, smaller ganglion cells regenerated better than large neurons by a significant 13.8%. These results indicate that the RE is not an obstacle for the re-growth of sensory fibers, but partially hinders fiber regeneration from motoneurons. They also suggest that fine fibers may be at an advantage over large ones to regenerate through the RE.