A 20 year (1963 to 1982) surgical experience including 175 consecutive patients with aortic dissections was analyzed by logistic discriminant analyses to identify predictors of high operative risk. The patient population had characteristics similar to those in large autopsy series. Sixty-nine percent had type A and 58% had acute dissections. The intimal tear was located in the ascending aorta in 60% of the patients, the descending aorta in 27%, and the transverse arch in 13%. The overall operative mortality rate was 23 +/- 3%. The operative mortality rates were substantially lower between 1977 and 1982: mortality in patients with acute type A dissections, 7 +/- 5%; in those with chronic type A, 11 +/- 7%; in those with acute type B, 13 +/- 12%; and in those with chronic type B, 11 +/- 11%. After preliminary univariate screening, the following factors were determined to be significant independent predictors of operative mortality (in rank order of declining predictive power): type A patients (n = 121), renal dysfunction, tamponade, renal/visceral ischemia, and operative date; type B patients (n = 54), rupture, renal/visceral ischemia, and age; all patients (n = 175), renal dysfunction, renal/visceral ischemia, site of tear (ascending less than descending less than arch), tamponade, operative date, and pulmonary disease. Interestingly, several variables had no important bearing on operative mortality, including type (acute vs chronic) of dissection, age, previous operation, rupture, stroke, paraplegia, Marfan's syndrome, concomitant aortic valve replacement and/or coronary artery bypass grafting, site of tear, and whether or not the tear was resected in type A patients; emergency operation, hypertension, previous cardiac symptoms, paraplegia, site of tear, and resection of tear in type B patients; and, when all patients were considered together, age, sex, cardiac symptoms, prior operation, stroke, paraplegia, acute myocardial infarction, acute aortic regurgitation, Marfan's syndrome, and tear resection. These data allow calculation of any individual patient's operative risk and document that the operative mortality rate today is relatively low for all patients with aortic dissections, irrespective of type or acuity. Earlier surgical referral of patients with acute type A or acute type B dissection before irreversible major end-organ ischemia and/or infarction is probably in part responsible for the substantially improved results since 1977.