Increased left ventricular (LV) mass is associated with increased cardiovascular morbidity and mortality. LV mass is commonly estimated from echocardiography according to the Penn or ASE (American Society of Echocardiography) conventions. No formal statistical test of agreement between these methods has been published. Therefore we compared M-mode echocardiographic LV mass estimates by the Penn and ASE methods in a normal adult population. M-mode echocardiographic tracings were obtained in 169 healthy volunteers and used to calculate LV mass using the Penn and ASE methods. Median values of the estimates were similar [Penn, 126 g (interquartile range 96-170 g); ASE, 129 g (105-164 g); P=0.08] and were highly intercorrelated (r=0.98, P<0.0001). However, the Bland-Altman analysis of agreement revealed significant inconsistencies between Penn and ASE LV mass values. The difference between Penn and ASE values was correlated significantly with heart size (P<0.0001), such that, for small hearts, the Penn LV mass was lower than the ASE LV mass; in contrast, for large hearts, Penn estimates were greater than ASE values. In the upper 5% of the LV mass distribution, the median value for the Penn LV mass index was 132.4 g/m(2), compared with 116.5 g/m(2) for ASE values (2P=0.017). Thus the two most common methods of echocardiographic estimation of LV mass differ significantly at the upper and lower ends of the heart size distribution. These results have important implications for both cardiac research and clinical evaluation.