From June 1960 to September 1990, 1414 patients underwent repair of thoracoabdominal aortic aneurysms, of whom 112 (8%) had pulmonary complications requiring respiratory support with tracheostomy; subsequently 45 (40%) died in the hospital. We determined by stepwise logistic regression analysis, in a prospective study of high-risk type I and II thoracoabdominal aortic aneurysms repairs, the independent predictors of respiratory failure, defined as respiratory ventilation exceeding 48 hours after operation. In 98 patients studied, 38 (39%) were women, 60 (61%) were men, 54 (55%) had type II thoracoabdominal aortic aneurysms, 34 (35%) had aortic dissection, 19 (19%) were nonsmokers, 40 (41%) exsmokers, and 39 (40%) active smokers. Before operation, 55 (56%) had chronic pulmonary disease with respiratory failure developing in 58% (p = 0.0005 versus no chronic pulmonary disease, 10/43, 23%), and of the 26 patients in the lower quarter of forced expiratory volume (1 sec) (FEV1 less than or equal to 1.45 L) respiratory failure developed in 61% (p = 0.035). In-hospital survival was 98% and 83% (p = 0.008), respectively, and cumulative survival at 6 months by Kaplan-Meier analysis was 96% and 80% (p = 0.004, log-rank test), respectively, for patients without respiratory failure (N = 56/98, 57%) and with respiratory failure (N = 42/98, 43%). On univariate analysis, the following were associated with respiratory failure (p less than 0.05): FEV1, FEV1% predicted, FVC, FEF25, FEF25% predicted, FEF25-75, FEF25-75% predicted, PaCO2 Pao2, symptoms, smoking history, chronic pulmonary disease, cryoprecipitate volume, postoperative neuromuscular deficit, cardiac complications, reoperation for bleeding, renal complication, stress ulceration, postoperative creatinine level, postoperative dialysis, and postoperative encephalopathy. The independent predictors of respiratory failure were (p less than 0.05): chronic pulmonary disease, smoking history, cardiac and renal complications. In patients with chronic pulmonary disease, the only independent predictor was FEF25 (p = 0.030). These observations may be of value in selecting patients for elective operation.