Previous studies to determine the sensitivity of the electrocardiogram (ECG) for left ventricular hypertrophy (LVH) in children had their imperfections: they were not done on an unselected hospital population, several criteria used in adults were not applied to children, and obsolete limits of normal for the ECG parameters were used. Furthermore, left ventricular mass (LVM) was taken as the reference standard for LVH, with no regard for other clinical evidence. The study population consisted of 832 children from whom a 12-lead ECG and an M-mode echocardiogram were taken on the same day. The validity of the ECG criteria was judged on the basis of an abnormal LVM index, either alone or in combination with other clinical evidence. The ECG criteria were based on recently established age-dependent normal limits. At 95% specificity, the ECG criteria have low sensitivities (<25%) when an elevated LVM index is taken as the reference for LVH. When clinical evidence is also taken into account, the sensitivity improved considerably (<43%). Sensitivities could be further improved when ECG parameters were combined. The sensitivity of the pediatric ECG in detecting LVH is low but depends strongly on the definition of the reference used for validation.