Drug-induced delay in ventricular repolarization and proarrhythmias have attracted considerable regulatory attention. The measure of delayed ventricular repolarization most frequently used clinically is the ability of the new chemical entity (NCE) to prolong the QTc interval on surface electrocardiogram. Before they can be approved, new chemical entities with systemic bioavailability require characterization for their potential to prolong the QTc interval. Inevitably, QTc interval prolongation has come to be recognized as a surrogate marker of the risk of torsade de pointes (TdP)--a unique form of potentially fatal polymorphic ventricular tachycardia. Although it is the best and the simplest clinical measure that is available at present, QTc interval is not a reliable surrogate of TdP. Intramyocardial dispersion of repolarization appears to play a more important role both in electrical stability of the ventricles and in arrhythmogenesis. The potential importance of myocardial dispersion of refractoriness in arrhythmogenesis has led to a number of attempts to assess it from the surface electrocardiogram. This review summarizes the evidence for and against the predictive value of one of these attempts-measurement of the so-called QT dispersion. Although the concept of QT dispersion is the best known and most widely investigated, it has also proved to be the least successful in predicting the risks of drug-induced TdP.