The 12-lead scalar electrocardiogram has been correlated with angiographically determined left ventricular dimensions in 103 consecutive patients in order to determine the reliability of the electrocardiographic diagnosis of left ventricular hypertrophy. Simple and multiple correlations between various parameters of QRS complex voltage and left ventricular mass, volume and wall thickness were poor but statistically significant. Mean voltage of QRS complex parameters was higher in patients with increased left ventricular mass and volume than in those with normal mass and volume, but the sensitivity of individual voltage parameters in identifying increased mass and volume was poor. A negative T wave in lead l or V6 was the most sensitive indicator of increased left ventricular mass and volume, but the specificity of this variable was poor. The sensitivity of "high voltage" of the QRS complex was increased (67 percent true positives) when moderate to marked increase in left ventricular mass had occurred, but at the expense of decreased specificity (13 percent false positives). It is concluded that while an increase in left ventricular dimensions tends to be associated with an increase in QRS voltage and a repolarization abnormality, the currently used electrocardiographic criteria for the diagnosis of left ventricular hypertrophy are both insensitive and nonspecific.