Calculation of left ventricular mass by M mode echo is based on the assumption that the geometry of the left ventricle is an ellipsoid, the long axis of which is twice that of its short axis. The hypothesis in not always true and often leads to overestimation of the ventricular mass. The authors propose a method combining M mode data (end diastolic dimension, septal and posterior wall thickness) and 2D measurement of the left ventricular long axis: the left ventricular mass was measured by Devereux's and the authors' methods in 185 hypertensives. The 2D measurement of the long axis (mean: 84.7 mm) was much smaller than twice the short axis (mean: 52.3 mm) and the two measurements were poorly correlated. Measurement of the long axis was reproducible. The two methods of calculation were closely correlated (r = 0.95) but, on average, 23% lower with the authors' method. These results seem to be more closely related to ambulatory blood pressure than those of the classical method. The authors' combined method takes into account the true geometry of the left ventricle better than M mode method alone and avoids overestimation of left ventricular mass and the prevalence of excentric left ventricular hypertrophy in hypertensive patients.