Since the original description of composite replacement of the aortic valve and ascending aorta by Bentall in 1968, several modifications of the technique have been described. In order to evaluate the results of these different techniques, we have retrospectively reviewed our results with 140 consecutive patients who underwent Bentall operations between October 1986 and March 1994, using three different anastomotic techniques: Classic, n = 30; Button, n = 95, and Cabrol, n = 15. Overall hospital mortality was 5%. In univariate analysis, acute type A dissection, rupture, new preoperative neurological symptoms, and the Cabrol technique were associated with a higher hospital mortality, but by multivariate analysis no independent risk factors were demonstrated. Overall rates of reoperation did not differ among the three techniques (Classic 4.1%/pt-yr, Button 2.7%/pt-yr, Cabrol 0%/pt-yr; p = 0.44). The actuarial freedom from reoperation was 87% at 5 years. The 5-year actuarial survival for all patients was 79% (Classic 85%, Button 82%, Cabrol 52%): the poorer results with the Cabrol modification are likely due to patient selection, complicated by a higher early mortality in this small group of patients. The presence of dissection was associated with a higher mortality in Marfan patients (50% vs 8%, p = 0.03). The rate of aortic valve-related complications was 3.6%/pt per year. Actuarial event-free survival was 67% at 5 years. Current indications for an elective Bentall procedure include an ascending aortic diameter of 6 cm or greater, with significant aortic valvular dysfunction, and dilatation of the ascending aorta greater than 5 cm in patients with Marfan syndrome or a bicuspid aortic valve. The routine procedure of choice is the Button Bentall technique, with the Classic Bentall and the Cabrol variation reserved for use under special circumstances.