A long term evaluation of prosthetic replacement of the aortic valve by various types of prosthesis has been carried out; 530 patients from various centres of cardiology were studied. The average annual mortality after the first two postoperative years was about 2.5% and half of the patients undergoing surgery were still alive after 10 years. The functional improvement is usually marked and is generally to be expected, although with an increased risk, even in patients who underwent surgery at stage IV of the NYHA classification. In this study, the prostheses which were associated with the best long term results were the Starr-Edwards 1260, the Smeloff-Cutter, and the Björk. The commonest complications after the first month were dehiscence of the prosthesis (11.8%), thromboembolic episodes (11.2%), coronary complicatons (11.1%) and cardiac failure (7.1%). The most serious is infection of the prosthesis, which is almost invariably fatal. Deterioration in the substance of the prosthesis has become very rare, but there must be reservations in this respect with the SE2400, which has still not been in use long enough. Revision operations were carried out on 7.6% of the survivors at one month, with a mortality of 18.4%. Sudden death remains a constant threat. A study of the actuarial survival graphs for operated and non-operated cases shows that surgery should be advised in all poorly tolerated valve defects, and even for an uncomplicated tight aortic stenosis. Bearing in mind the progressive lowering of operative mortality, surgical treatment is also justified in cases with marked or gross aortic incompetence which is well tolerated if flow-up studies show a progressive increase in cardiac volume.