Objective To observe the effects of early-stage exercise on function recovery and ED1 expres-sion after spinal cord injury. Methods Twenty-two SD rats were randomly divided into a control group ( n=8) , an exercise group ( n=8) and a sham-operated group ( n=6) . The control and exercise groups received T9 spinal cord contusions. The sham-operated group received only T9 laminectomy without spinal cord injury. Two rats died within 6 hours after the operation and 3 others were abandoned because of motor function recovery just after the spi-nal cord contusion. The remaining 17 were included in the statistical analysis: a control group of 6, an exercise group of 5 and a sham-operated group of 6. BBB scoring was conducted 2 days later and then weekly until 8 weeks after the operation. The exercise group began treadmill and wheel running exercise on day 2 and continued 5 times per week until the 8th week. After paraformaldehyde perfusion, fixation and embedding, frozen spinal cord slices were Nissl stained for ED1 and GFAP immunofluorescence measurements. Results Compared with the control group, the average BBB score of the exercise group was significantly higher at each week except weeks 2 and 4. By weeks 7 and 8, the exercised rats showed 13.60 and 14.60 BBB punctuation, which means that those animals took frequent, consistent, weight-supporting plantar steps, consistent with FL-HL coordination. That motor behavior was not achieved by the control group even occasionally. In terms of the tissue morphology and immunohistochemical staining results, the empty area was not significantly reduced in the exercise group compared with the controls, but the damaged area was significantly reduced. In addition, ED1 positive cells in the lesioned area were significantly reduced in the exercised group compared with the controls. Conclusions Exercise can promote the recovery of motor function after spinal cord injury. The mechanism may involve reducing the lesioned area and inflammatory response.