BACKGROUND Mechanical prosthesis is the choice of valve at the mitral position in children, although re-replacement of prostheses because of prosthesis-patient mismatch is almost inevitable when prostheses were implanted in small children. The methods to predict prosthesis-patient mismatch as a result of patients' somatic growth or pannus formation in children by noninvasive methods have not been well established. METHODS Thirty-two children underwent mitral valve replacement with 37 bileaflet mechanical prostheses (26 St. Jude Medical prosthetic valves, and 11 CarboMedics prosthetic valves) and were followed up a mean of 6.8 years (maximum 18.3 years) with a complete follow-up rate of 94%. RESULTS There were no operative deaths and 5 late deaths. Re-replacement of mitral valve because of prosthesis-patient mismatch was required in 5 patients. Freedom from valve-related events and re-replacement of mitral valve at 15 years were 32% +/- 23% and 54% +/- 18%, respectively. Actuarial survival rate was 63% +/- 19% at 15 years. Prosthetic valve orifice area index (manufactured geometric prosthetic valve area divided by patient's body surface area) was well correlated with maximum transprosthesis flow velocity estimated by Doppler echocardiography during follow-up, whereas valve orifice area index had no significant correlation with pulmonary artery wedge pressure assessed by cardiac catheterization. Maximum transprosthesis flow velocity had a significant correlation with pulmonary artery wedge pressure. CONCLUSIONS Valve orifice area index itself was not a reliable index to predict prosthesis-patient mismatch. Maximum transprosthesis flow velocity was a useful index to predict pulmonary artery wedge. Invasive cardiac catheterization to determine re-replacement of the prosthesis should be considered when maximum transprosthesis flow velocity exceeds 270 cm/s.