Emerging evidence indicates that etiology plays an essential role in the results of mitral valve repair. In an attempt to evaluate the possibilities and problems of repair in a rheumatic population, all consecutive patients treated between July 1988 and July 1990 were reviewed. Three hundred and four patients at risk of undergoing a repair were studied. The mean age was 31.3 years. In 78 patients (25.6%) the valve was considered beyond repair and immediately replaced (MVR). In 26 patients (8.5%) an unsuccessful attempt at repair was followed by replacement (attempts), and 200 (65.7%) underwent a successful repair (repair). The repair group was younger (p less than 0.001) and had less tricuspid involvement (p less than 0.005). The repair rate was highest in regurgitant lesions (79.4% repaired, p less than 0.005). The attempts group was very similar to the MVR group. The total hospital mortality was 3.2%. For MVR it was 3.8%, for attempts 19.2%, and for repairs 1.0%. The incidence of embolism was 3.34%/patient/yr for the mechanical, 3.06%/patient/yr for bioprosthesis, and 2.19%/patient/yr for repairs. Only 7.5% of repairs were anticoagulated. A reoperation was needed in 26 patients: one for bioprosthesis, one for aortic repair failure, and 24 for mitral repair dysfunction (12.1% of all surviving repairs). The reoperation incidence after repair was 26.8% in patients with an age under 20 years and 4.5% in those over 20. Eight patients had active carditis and 10 severe hemolysis. There were eight late deaths (six prosthesis and two repairs), with an actuarial total survival at 30 months of 72.46 +/- 8.45% for prosthesis and 94.41 +/- 3.74% for repairs (p less than 0.005).(ABSTRACT TRUNCATED AT 250 WORDS)