We compared the performance of precordial QT dispersion, late potentials on the signal-averaged electrocardiogram (ECG), and reduced left ventricular ejection fraction for identification of inducible ventricular tachycardia (VT) in 162 patients undergoing electrophysiologic study (EPS). QT(apex) dispersion in 56 patients with inducible VT (72 +/- 55 msec) was greater than that in 106 patients without inducible VT (55 +/- 36 msec, p < 0.01); dispersion was greater in both groups than in 144 normal subjects (33 +/- 19 msec). A QT(apex) dispersion partition of more than 68 msec, the upper ninety-fifth percentile in normal subjects, identified inducible VT with a specificity of 75% and a sensitivity of 45%. Although the performances of late potentials (specificity 82%, sensitivity 59%) and reduced ejection fraction (specificity 86%, sensitivity 54%) were each stronger than QT dispersion alone for identification of inducible VT, abnormal QT(apex) dispersion remained a significant additional predictor of inducible VT in a logistic regression model that included the three variables (specificity 78%, sensitivity 75%).