Objective To compare the surgical efficacy of different operating methods for treating old thoracolumbarfracture with spinal cord injury. Methods From September 2000 to March 2006, 34 cases of old thoracolumbar fractures with spinal cord injury were treated. Patients were divided into 2 groups randomly. Group A (n=18): anterior approach osteotomy, il iac bone graft and internal fixation were used. There were 10 males and 8 females with the age of 17-54 years. The apex level of kyphosis was T11 in 2 cases, T12 in 5 cases, L1 in 8 cases and L2 in 3 cases. The average preoperative Cobb angle of kyphosis was (36.33 ± 3.13)°, and the average preoperative difference in height between anterior and posterior of involved vertebra was (22.34 ± 11.61) mm. Neurological dysfunction JOA score was 10.44 ± 1.12. Group B (n=16): transpedicular posterior decompression and internal fixation were used. There were 8 males and 8 females with the age of 18-56 years. The apex level of kyphosis was T11 in 2 cases, T12 in 6 cases, L1 in 7 cases and L2 in 1 case. The preoperative Cobb angle of kyphosis was (38.55 ± 4.22)°, and the preoperative difference in height between anterior and posterior of involved vertebra was (20.61 ± 10.22) mm. Neurological dysfunction JOA score was 10.23 ± 2.23. Results All the patients were followed up for 9-46 months with an average of 13.5 months. Cobb angle was (12.78 ± 3.76)° in group A, which was improved by (24.23 ± 1.64)° campared to that of preoperation; and was (10.56 ± 4.23)° in group B, which was improved by (26.66 ± 1.66)°. JOA score was 14.21 ± 1.08 in group A, which wasimproved by 3.92 ± 1.33; and it was 13.14 ± 2.32 in group B, which was improved by 3.12 ± 1.95. The average postoperative difference between anterior height and posterior height of vertebral body in group A was (3.11 ± 1.06) mm, which was improved by (18.03 ± 2.14) mm; and it was (2.56 ± 1.33) mm in group B, which was corrected by (20.36 ± 3.78) mm. There were statistically significant differences in the above indexes between preoperation and postoperation in 2 groups (P lt; 0.01), but no significant differences between 2 groups (P gt; 0.05). In group A, pleural effusion occurred in 2 cases and local pulmonary collapse in 4 cases and intercostals neuralgia in 1 case. In group B, leakage of cerebrospinal fluid occurred in 3 cases. Conclusion Both anterior and posterior approach are capable of treating of the old thoracolumbar fracture with incomplete spinal cord injury and providing the satisfying result of deformation correction, neurological decompression and neurological functional recovery to a certain extent.