Aortic dissection (AD) is a disease with a high-risk of mortality. Late deaths are often related to complications in nonoperated aortic segments. Between 1984 and 1996, we retrospectively analyzed the data of 109 patients with acute AD (81 men and 28 women; average age 61 +/- 14 years). All imaging examinations were reviewed, and a magnetic resonance imaging examination was performed at the time of the study. Aortic diameters were measured on each aortic segment. Predictive factors of mortality were determined by Cox's proportional hazard model, in univariate and multivariate analyses, using BMDP statistical software. Follow-up was an average of 44 +/- 46 months (range 24 to 164). Actuarial survival rates were 52%, 46%, and 37% at 1, 5, and 10 years, respectively, for type A AD versus 76%, 72%, and 46% for type B AD. Predictors of late mortality were age >70 years and postoperative false lumen patency of the thoracic descending aorta (RR 3.4, 95% confidence intervals 1.20 to 9.8). Descending aorta diameter was larger when false lumen was patent (31 vs 44 mm; p = 0.02) in type A AD. Furthermore, patency was less frequent in operated type A AD when surgery had been extended to the aortic arch. Thus, patency of descending aorta false lumen is responsible for progressive aortic dilation. In type A AD, open distal repair makes it possible to check the aortic arch and replace it when necessary, decreases the false lumen patency rate, and improves late survival.