In spinal cord injuries, various attempts have been made to reconstruct neural connections once disrupted. To improve current procedures and develop therapeutic methodologies it appears important to compare these reconstructive attempts via a standardized quantification of any ensuing functional recovery with parallel correlations to any potentially repaired neural connections. We have reported previously a quantitative assessment of neural connections across the graft site of rats whose spinal cord segments were neonatally replaced with embryonic spinal cord segments or a peripheral nerve section under comparable conditions. Using this same experimental model the present study assessed locomotor performance quantitatively using an open field locomotor scale at various postoperative intervals from day 0 to 5 weeks postinjury. To examine hind-forelimb coordination in further detail, electromyography was employed to record simultaneously from all four limbs during locomotion. Half of the rats whose spinal cord segments were repaired by replacement with embryonic homologous structures acquired virtually normal locomotor function, with a delay of five days compared with that of sham-operated rats. Detailed analysis revealed an abnormality in ankle joint movement and the stability of trunk during locomotion. Electromyography revealed that the pattern of locomotion in these rats was similar to controls. Grafted segments joined with the host spinal cord without gliosis at the host-graft interface. The remaining rats with an embryonic tissue graft showed various grades of hind-forelimb coordination. Gliosis and cavity were observed at the host-graft interface. The rats whose spinal cord was repaired by periphral nerve graft lacked hind-forelimb coordination despite the achievement of weight-supported steps. It appears likely that the grade of locomotor performance depends on quantity and quality of reestablished neural connections across the graft.