Twenty-two consecutive patients with thoracic aortic disruptions were treated over a three-year period. With the exception of 3 patients who required emergency thoracotomy, aortography was performed on all patients. The disruption was identified just distal to the left subclavian in all cases. Two patients died intraop eratively prior to repair. A Gott shunt was utilized in 5 patients, cardiopulmonary bypass in 1, and the "clamp-and-sew" technique in 14. Paraplegia occurred in 2 patients (14%) of the "clamp-and-sew" group. These patients had aortic cross- clamp times in excess of thirty minutes. Paraplegia did not develop when either cardiopulmonary bypass or a Gott shunt was used. Additionally, multiple tears of the descending aorta, which were not visualized on aortography, were found intraoperatively in 2 of 19 patients (10.5 % ). One was treated with graft insertion on cardiopulmonary bypass, and the second, with graft insertion of the "clamp- and-sew" technique. The second patient developed paraplegia, attributed to the prolonged clamp time. In conclusion, a shunt procedure would seem to provide better protection of the spinal cord, especially when multiple sites of aortic injury are identified.