In this review the present state of cardiac valve replacement is summarized on the basis of the literature and personal experience in approximately 1500 cases, using various prosthetic models. Durability, hemodynamic performance at rest and during exercise, and thromboembolic risk are considered in particular and our choice of the various valve models is discussed on these grounds. The long-term durability has been significantly improved and has possibly been solved in the more advanced mechanical valves, while this question is still open in the present biological prostheses. Hemodynamic performance is considered less than optimal in all conventional mechanical prostheses and in the xenografts. Pressure gradients are abnormal in all mechanical as well as in stented biological prostheses, the latter being less likely to solve the problem of a narrow aortic root. The significantly lower incidence of thromboembolism and total valve thrombosis and the absence of anticoagulation accidents in patients with bioprostheses has tipped the balance for the present in favor of the latter, particularly in mitral valve replacement and certainly in all those patients in whom anticoagulation in impossible, unlikely, or undesirable.