The role of surgery in the management of acute spinal cord or cauda equina injuries remains controversial. The present study analyzed ten admission features and three outcome variables in 208 patients treated in an Acute Spinal Cord Injury Unit, 116 (56%) of whom underwent at least one spinal operation. The surgical and non-surgical groups showed no significant differences in the following seven clinical features: age, sex, distance travelled to the Unit, time interval between trauma and admission, type of accident, severity of injuries to the spinal cord, and severity of associated injuries. However, the two groups showed significant differences in level and type of vertebral column injury, and in the frequency of pre-existing spinal abnormalities. These differences were due to management policies which selected certain injuries for surgical or non-surgical treatment. One-third of the operative procedures were performed primarily for neural decompression, one-third primarily for reduction of bony structures and one-third for fusion. However, 95% of the operative patients had a fusion at the initial operation. Operative treatment was associated with a lower overall mortality rate (6.1%) than non-operative (15.2%), despite a higher frequency of thrombo-embolic complications in the surgical group. Overall, there was no difference between operated and non-operated patients in length of stay or neurological recovery. Surgical management of patients with acute spinal cord injury appears safe in terms of mortality rate and neurological recovery, but it has not been proven to improve the latter.