Between June 1985 and June 1991, 63 children underwent surgical reconstruction of the mitral valve for rheumatic (46 cases), congenital (12 cases) and myxoid (five cases) disease. The ages ranged from two to 18 years (mean 14.1±3.7 years). Valvar dysfunction was classified according to its pathophysiological abnormalities. A group of four cases presented with regurgitation secondary to lesions located in the valvar structures but with normal motion of the leaflets. In a second group of 14 cases, mitral regurgitation was due to prolapsed leaflets because of lesions located mainly in the subvalvar structures. A third group was formed by 35 patients with mitral regurgitation with restricted motion of the leaflets due to lesions in the valvar and subvalvar structures. The final group, of 10 cases, presented with mitral stenosis. The overall surgical mortality rate was 4.7% (3/63), and follow-up data were available in all survivors from one to 96 months (mean 33.4±25.4). Four cases underwent reoperation due to residual incompetence, one case due to bacterial endocarditis, and two more are scheduled for replacement of the valve due to unfavorable evolution, giving an overall rate of reoperation of 4.3% per patient/year. One patient died in the period following valvar replacement (late mortality rate of 0.6% per patient/year). Thromboembolism occurred in four cases in the absence of anticoagulation; three of them were in atrial fibrillation (late thromboembolic rate 2.4% per patient/year). Prior to surgery, 28 cases were in functional class II of the New York Heart Association, 34 patients were in class III and one patient in class IV. At the end of the follow-up period, 49 patients were in class I, seven in class II and four in functional class III (p<0.0001). The cardiothoracic ratio before surgery ranged from 0.40 to 0.81 (mean 0.60±0.07) and, after surgery, the values ranged from 0.40 to 0.79 (mean 0.55±0.07) (p<0.0001). Randomized late echocardiographic evaluation in 24 cases revealed residual mild mitral regurgitation in 20 cases, moderate in two and severe in two. The latter are scheduled for valvar replacement. There were no significant differences in the surgical results among the four groups. Reconstruction of the mitral valve, therefore, provides stable functional results with low surgical and late mortality, as well as an acceptable rate of reoperation irrespective of the lesions of the valvar apparatus.