Prediction of the onset of Torsade de Pointes (TdP) is a challenge for clinicians, because the list of drugs affecting myocardial repolarisation is continuously increasing. Alterations in the activity of autonomic nervous system and abnormalities in ventricular repolarisation are key features both as triggers and as markers for vulnerability to TdP. Recent molecular genetic studies have shown that autonomic nervous system has channel and gene specific influences on vulnerability to TdP. New analysis techniques in quantifying the dispersion of repolarisation have also been developed. QT interval dispersion, defined as a difference between the maximum and minimum QT interval measured from the standard 12-lead electrocardiogram (ECG), is one such method. In preliminary studies, QT dispersion has provided more accurate information on the risk for TdP than the measurement of the length of QT interval from a single ECG lead. Unfortunately, QT dispersion is entailed with some conceptual and methodological problems, which impairs its widespread clinical utility in risk stratification. Despite advances in the understanding of the role of autonomic nervous system as a trigger of TdP in specific gene mutations and improved clinical methods in detecting repolarisation abnormalities, accurate and reliable prediction of the onset of TdP still remains an unresolved clinical problem in individual cases.