Objectives: In severe mitral regurgitation (MR), there are few reliable parameters about left ventricular (LV) function or performance. LV stroke work index (LVSWI), an invasively-acquired parameter, may provide more relevant information about cardiac performance in severe MR. Our aim was to measure LVSWI in severe MR noninvasively with echocardiography and validate its clinical usefulness. Methods: We evaluated 32 patients with severe MR prospectively, whose mean age was 55.3 짹 11.2 years. Cardiac catheterization with Swan-Ganz catheter was performed, followed by echocardiographic evaluation in each patient. Thermodilution method was used for the measurement of invasive LVSWI, whereas echocardiographic measurement of stroke volume (SV) at LV outflow tract (LVOT) and LVEDP estimated by the ratio of peak early diastolic mitral inflow velocity (E) and peak early diastolic mitral annular velocity (E���) - E/E���- were used for evaluation of noninvasive LVSWI. Symptom-limited exercise treadmill test with expired gas analysis was performed and peak oxygen consumption rate (VO_{2 max}) and plasma B-type natriuretic peptide (BNP) was measured. Intraclass correlation coefficients were measured and linear regression analysis was performed between parameters. Results: Mean value of cardiac output, cardiac index and SV measured by thermodilution methods were 4.4 짹 1.4 L/min, 2.6 짹 0.9 L/min/m² and 57.9 짹 22.4 cc, respectively. E/E��� was well correlated with LVEDP (r=0.576, p=0.001) and LVEDP could be estimated by the following formula, which is LVEDP=5.32+0.61x(E/E���). The mean value of noninvasive LVSWI (calculated from SV at LVOT) was 32.7 짹 10.3 g쨌m/m², which correlated well with invasive LVSWI (mean 짹 SD; 34.2 짹 14.9 g쨌m/m², intraclass correlation coefficient= 0.848, p<0.001). This noninvasive parameter was well correlated with VO_{2 max} (r=0.535, p=0.003) and plasma BNP level (��=0.583, p=0.001). Conclusions: Echocardiographically-estimated LVSWI may be a reliable parameter for evaluating true LV performance in severe MR.