We evaluated the long-term outcome of mitral valve replacement with a Harken caged-disc prosthesis for up to 11 years (range, 50 to 130 months; mean, 81 months) in 170 patients whose mean age was 55 years. The early (30-day) mortality was 11.2% (19 out of 170 patients). Late follow-up information was obtained for 144 (95%) of the 151 survivors. The actuarial survival was 57% at 5 years and 40% at 10 years. Overall mortality was associated with advanced age, male sex, an ischemic origin for the mitral valve disease, and nonuse of warfarin anticoagulation. Late deaths (n=59) were valve-related in 46%, cardiac but non-valve-related in 44%, and noncardiac in 10% of the cases. One thromboembolic event or more occurred in 41 patients (incidence, 5.7% per patient year), 14 of whom died (24% of the late deaths). All four patients who were not on warfarin, aspirin, or other antithrombotic therapy experienced thromboemboli. This complication was correlated with the nonuse of warfarin-type anticoagulation, with mitral regurgitation, and with late cardiac death. Mechanical prosthetic failure resulted in reoperation or death in 7.6% of the late survivors (1.5% per patient year). In 75 patients with normally functioning prostheses, the disc-to-sewing ring ratio was established by means of cinefluoroscopy (0.93 +/- 0.04, mean +/- 25D). Because of the high proportion of cardiac valve-related deaths (46%), the high incidence of late mortality due to thromboembolic events (24%), and the 7.6% incidence of reoperation or death resulting from mechanical valve failure, close follow-up with cinefluoroscopy and continued warfarin anticoagulation (alone or in combination with dipyridamole) are essential after mitral valve replacement with the Harken caged-disc prosthesis.