Between January 1989 and July 1992, 76 patients with thoracolumbar fractures were operatively treated at the Department of Trauma Surgery, Hannover Medical School. After a mean of more than 3 years, 56 of 62 patients (90%) still alive who had their implants removed were examined. According to the ASIF classification 33 patients sustained type A fractures, 13 type B and 10 type C. Three patients with incomplete paraplegia returned to normal; in one case of complete paraplegia no change occurred. In 40 cases the dorsal instrumentation was combined with transpedicular cancellous bone grafting. The mean operative time totaled 3 h. In this series, two complications (3.6%) were observed: one iatrogenic vertebral arch fracture without consequences and one deep infection. Compared to the preoperative status, our follow-up examinations demonstrated permanent physical and social sequelae; the percentage of individuals able to do physical labor was reduced by half (22 to 11 patients) whereas the share of unemployed or retired patients doubled (4 to 8 patients). At the time of follow-up examination only 21 of 42 patients continued in sports. The assessment of complaints and functional outcome with the "Hannover Spinal Trauma Score" reflected a significant difference (P < 0.001) between the status before injury (96.6/100 points) and at the time of follow-up (71.4/100 points). The radiographic assessment in the lateral plane (Cobb technique) demonstrated a significant (P < 0.001) mean restoration from an initial angle of -15.6 degrees (kyphosis) to +0.4 degree (lordosis). Serial postoperative radiographic follow-up showed progressive loss of correction; at follow-up examination we found a mean of 10.1 degrees (P < 0.001). Compared to the preoperative deformity a mean improvement of 6.1 degrees to an average of -9.7 degrees at follow-up examination was noted. The addition of transpedicular cancellous bone grafting did not decrease the loss of correction. CT scans after implant removal were performed in 9 cases: only 3 of 9 patients showed evidence of intervertebral fusion. No correlation could be found between ASIF classification and radiographic outcome. However, the preoperative wedge angle of the vertebral body correlated significantly with the postoperative loss of reduction. Due to disappointing results after dorsal stabilization with transpedicular cancellous bone grafting we recommend a combined procedure with dorsal stabilization and ventral fusion in cases of complete or incomplete burst injury of the vertebral body.