Operative correction of ischemic mitral regurgitation (IMR) is associated with high risk approach. The objective of this retrospective study was to examine the interaction between the various underlying pathophysiologic mechanisms, the operative procedure, and their influence on short- and long-term outcomes. Over a 10-year period starting January 1984, mitral valve repair or replacement was performed on 150 patients with IMR. The age range was 42-86, mean 67, years; 71 (47%) were females; 139 (93%) were in NYHA functional class III or IV; 23 (15%) were reoperations; and 30 (20%) were in atrial fibrillation. Functional IMR due to annular dilatation or restrictive leaflet motion was present in 106 (71%), and structural IMR due to ruptured chordae or papillary muscle in 44 (29%). Mitral valve repair was performed in 94 (63%) with an annuloplasty ring employed in 80 (85%) patients. Mitral valve replacement was performed in 56 (37%), with 40 (71%) receiving a bioprosthesis (32 Hancock and 8 Carpentier-Edwards valves) and 16 (29%) a St. Jude valve. Coronary artery bypass graft surgery was performed in 139 (93%) patients. The overall operative mortality (OM) was 14/150 (9.3%). The OM for repair was 9.5% compared to 8.9% for replacement (P = NS). There was higher OM in the elderly, particularly in the repair group (P = 0.053), and a trend towards reduced OM in the recent years of the study (P = NS). No predictors of OM were identified by multivariate logistic regression analysis. Long-term follow-up was 98% complete and ranged from 2-120, mean 31.2, months for a total of 935 patient-years. The overall 5-year survival rate was 71 +/- 6%, with 91 +/- 5% for the replacement group compared to 56% +/- 10% for the repair group (P = 0.01). The functional subset of IMR who had a repair had the worse long-term survival (43 +/- 13%) compared to the structural/repair (76 +/- 13%) and structural/replacement groups (89 +/- 8%), and 92 +/- 7% for the functional/replacement group ((P = 0.0049). Multivariate logistic regression analysis identified the functional/repair group (hazards ratio 4.4; +/- 95%, confidence interval 1.6, 11, (P = 0.0031); and earlier years of surgery (hazards ratio 4.7; +/- 95% confidence interval 1.021; (P = 0.046) to be predictors of worse long-term survival. These results suggest that, in IMR, the underlying responsible pathophysiologic mechanisms appear to be the major determinants of survival, rather than the choice of the operative procedure.