The continued good results after mitral valve reconstruction prompted this retrospective study to compare operative and late results from our institutional experience since 1976 with 975 porcine mitral valve replacements (MVRs) (1976 to December 1987), 169 mechanical MVRs (1976 to December 1987), and 280 Carpentier-type mitral valve reconstructions (CVRs) (1980 to mid-1988). The operative mortality was 2.0% for isolated CVR, 6.6% for isolated mechanical MVR, and 8.5% for isolated porcine MVR. The overall operative mortality was 5.0% for CVR, 16.6% for mechanical MVR, and 10.6% for porcine MVR. The overall 5-year survival including hospital deaths was 76% for CVR, 72% for mechanical MVR, and 69% for porcine MVR. By multivariate analysis, the predictors of increased operative risk and of decreased survival were age, New York Heart Association functional class IV status, previous cardiac operation, and performance of concomitant cardiac surgical procedures. The type of valvular procedure was not predictive of operative risk or overall survival. The 5-year freedom from reoperation was 94.4% for nonrheumatic patients having CVR, 77.4% for rheumatic patients having CVR, 96.4% for mechanical MVR, and 96.6% for porcine MVR (p less than 0.05, rheumatic patients with CVR versus both MVR groups). The 5-year freedom from all valve-related morbidity and mortality was significantly better for valve reconstruction compared with both types of valve replacement. Thus, the operative risk and late survival obtained after mitral valve reconstruction were at least equivalent to those obtained after MVR. In addition, patients receiving mitral valve reconstruction had less valve-related combined morbidity than patients receiving valve replacement, thus making mitral valve reconstruction preferable in some patients with mitral insufficiency.