To assess the correlation among exercise tolerance, mitral valve are (MVA) and the left ventricular (LV) systolic performance, dynamic exercise echocardiography was performed on 21 patients with mitral stenosis (MS). The patients were divided into 3 groups by the attained work load and exercise time: 6 patients whose attained work load was 25 watts and exercise time was less than 6 minutes (group I); 7 patients, 50 watts and less than 9 minutes (group II); and 8 patients, 75 watts and 9 minutes (group III). MVA in group I was smaller than that in group III, however, the correlation between exercise time and MVA was poor (r=0.48). Resting LV end-diastolic and end systolic dimensions (EDD, ESD) and systolic functions (% shortening fraction, mean Vcf) showed no statistical differences among three groups. During exercise EDD decreased significantly in all three groups, but at peak exercise it did not show significant differences among three groups. ESD, whereas, decreased significantly with exercise in all three groups, but at peak exercise ESD in group I was significantly larger than that in group III (p<0.01). Per cent shortening fraction (% SF) decreased or unchanged with exercise in group I and in group II, while it increased significantly in group III (p<0.01). Mean Vcf in group I did not increase significantly with exercise, either. The extent of the increase in % SF and mean Vcf at peak exercise correlated well with exercise time (r=0.83, r=0.79). We conclude that exercise tolerance in MS patients is dependent not only MVA but also LV systolic function, and thus, exercise echocardiography is a useful method to evaluate the LV functional reserve in MS patients.