The surgical management of congenital mitral valve (MV) anomalies remains controversial and complex. Valve repair has been said to be preferable to valve replacement. To assess the validity of this suggestion we have reviewed our experience of these procedures at The Hospital for Sick Children, Great Ormond Street, London from 1969-1983. In that time 48 patients have undergone surgery; in 23 repair was practicable (Group A) and in 25 replacement of the MV was required (Group B). Group A patients were slightly older than group B patients (5.9 +/- 1.0 years (SEM) vs 3.6 +/- 0.5 years). Björk Shiley valves were the predominant prosthesis used. Overall mortality was 33%, with 17% in Group A and 48% in Group B. Hospital mortality was 4.5% in Group A and 28% in Group B. Actuarial survival for Group A was 90% at 1 year, and 75% at 2 and 5 years; for Group B it was 52% at 1 and 2 years and 42% at 5 years. Mortality was significantly higher in those aged less than 5 years. Four patients in Group A have required re-operation, in 3 valve replacement was required; 2 of these died in hospital. One patient in Group B has required reoperation for valve-related thromboembolism. Three patients are awaiting replacement of calcified bioprostheses. These data support the concept that repair of MV should be performed where possible. The severity of the anomalies was greater in Group B and a rethink as to whether valve replacement is the best option for severe mitral stenosis in those aged less than 5 years seems indicated, in view of the very high mortality.