QT dispersion (QTD) on 12-lead ECGs has been proposed as a marker of malignant ventricular tachyarrhythmias, and increased QTD has been reported in long QT syndrome (LQTS). On the other hand, it has been demonstrated that transmural dispersion is associated with ventricular tachyarrhythmias in an experimental model. However, the precise type of QTD or transmural dispersion that contributes most to ventricular tachyarrhythmias in patients with LQTS remains unclear. We evaluated 27 patients with acquired LQTS. These patients were divided into two groups: group A (n =12), patients with polymorphic ventricular tachycardia [torsades de pointes (TdP)], and group B (n =15), patients without TdP. The QT intervals were corrected using Bazett's formula. QTD was measured as the difference between the maximum and the minimum QT intervals, and T wave peak-to-end interval divided by the QT interval (Tpe) in the V5 lead was measured as a new index. Both the corrected QTD (QTDc) and Tpe were significantly larger in group A than in group B. Logistic regression analysis revealed that a reliable predictor for TdP in the QT variables in these patients was not QTDc but Tpe. Cumulative frequency distributions revealed that a Tpe of 0.28 is a good cut-off point for TdP. Tpe did not correlate with the corrected maximum QT interval, whereas the QTDc did correlate with this parameter. In conclusion, Tpe may be the best predictor for TdP in patients with acquired LQTS.