The ability to estimate left ventricular mass (LVM) from the standard 12-lead electrocardiogram (ECG) has been shown to be limited because there is a considerable variability of the normal 12-lead ECG due to demographic and anthropometric variables. We sought to study LVM in healthy subjects and its relationship with QRS duration, and established electrocardiographic criteria for left ventricular hypertrophy. Cardiac magnetic resonance imaging was used to measure LVM. Seventy-one healthy volunteers (36 men; age range, 21-82 years) were studied. All ECG criteria tested showed a statistically significant relationship with LVM. The highest R value was found between LVM and QRS duration, as well as the 12-lead voltage-duration product (R = 0.59, P < .001 for both). The lowest R value was found for the Sokolow-Lyon voltage criterion (R = 0.25, P = .033). Left ventricular mass differed significantly between sexes, as did all ECG criteria except the Sokolow-Lyon criterion. Thus, in healthy subjects, QRS duration alone is equally or more strongly correlated to LVM than are established electrocardiographic left ventricular hypertrophy criteria.