Between January, 1969, and January, 1978, 551 patients with mitral incompetence were treated by a system of reconstructive techniques. Mitral valve incompetence was classified into three types according to leaflet pliability; type I normal leaflet motion, 150 cases; type II, leaflet prolapse, 213 cases; and type III, restricted leaflet motion, 188 cases. Associated tricuspid valvular disease was present in 174 cases (31.5%) and was treated by prosthetic ring annuloplasty. The operative mortality rate was 4.2% (16/377) in the mitral group and 14% (25/174) in the mitral-tricuspid group. Follow-up data are available in 341 patients from 1 year to 10 years (average 4 1/2 years). The late mortality rate was 7% (24/341). Actuarial curves including hospital mortality rate show an 82% survival rate at 9 years in the mitral group and a 79% rate in the mitral-tricuspid group. Thirty-seven patients (11%) underwent reoperation mainly for residual (17) or recurrent (16) mitral incompetence. Thromboembolism occurred in 12 patients for an embolic rate of 0.6% per patient-year, even though 48% were not given anticoagulants. Acorrding to the New York Heart Association (N.Y.H.A.) classification, 76% (207/270) of the patients were in Class I, 19% (51/270) were in Class II, 4% (10/270) were in Class III, and 0.7% were in Class IV (2/270). Results of postoperative catheterization and angiocardiography are available in 52 patients. Comparison between the various groups shows that the best results were obtained in type II mitral incompetence, followed by type I and type III mitral incompetence. This experience demonstrates that predictable and stable long-term results have been achieved by techniques of valvular reconstruction with a low incidence of thromboembolism. Reproducibility of the techniques is a limiting factor which can be overcome by adequate training and progressive experience. Patient selection is based on the valvular disease rather than age, physical condition, or cause of valvular disease.