A metaanalysis of articles concerning the surgical management of acute traumatic rupture of the descending thoracic aorta published in the English-language literature between 1972 and July 1992 was performed. The overall mortality of 1,742 patients who arrived at the hospital alive was 32.0%, one-third died before surgical repair was started. Paraplegia was noted preoperatively in 2.6% of these hospitalized patients, and paraplegia complicated the surgical repair in 9.9% of 1,492 patients who reached the operating room in a relatively stable condition. Patients then were analyzed according to the surgical intervention used. Simple aortic cross-clamping (n = 443) was associated with a hospital mortality of 16.0% and incidence of paraplegia of 19.2%, despite lower average mean cross-clamp times (32 minutes; p < 0.01 versus passive or active methods of providing distal perfusion). In a subset of 290 patients in whom individual data were available, the cumulative risk of paraplegia was shown to increase substantially if the duration of aortic cross-clamping exceeded 30 minutes, but only when distal perfusion was not augmented (p < 0.00001). "Passive" perfusion shunts (n = 424) were associated with a mortality of 12.3%, and the incidence of paraplegia decreased to 11.1% (p < 0.001). However, shunts inserted from the apex of the left ventricle had a contradictory high 26.1% incidence of paraplegia compared with shunts from the ascending aorta (8.2%; p < 0.02).(ABSTRACT TRUNCATED AT 250 WORDS)