Eighteen years of continuous clinical use with the Starr-Edwards ball-valve prosthesis provides an extensive data base for comparison with more recently introduced cardiac prostheses, and establishes the credentials of this valve for current clinical use. Since 1965, 558 aortic (model 1260) and 285 mitral (model 6120) ball-valve prostheses have been inserted at the University of Oregon. Thromboembolic rates for the entire series (1965-1983) and our more recent experience (1973-1983) were 3.8% per year and 2.8% per year for the aortic valve; and 5.1% per year and 2.6% per year for the mitral valve. Patients who had experienced one embolic episode were substantially more likely to experience a subsequent event. Thrombotic stenosis was seen at rates of 0.1% per patient year and 0.4% per patient year for the aortic and mitral valves, respectively. These results are compared with other current series using a variety of mechanical and tissue prostheses. The Starr-Edwards valve remains a durable mechanical cardiac prosthesis with thromboembolic complications similar to most other current prostheses. Thrombotic stenosis tends to be rare, gradual, and electively managed as opposed to the sudden, unpredictable, and catastrophic thrombosis seen with tilting disc valves.