The identification of patients at high risk of sudden cardiac death is one of the greatest challenges for cardiologists. Non-invasive methods have, characteristically, low predictive sensitivities and specificities. The role of abnormalities of ventricular repolarisation (QT interval) in the genesis of ventricular arrhythmias has been well established by experimental data. For this reason, parameters of ventricular repolarisation on the surface electrocardiogram have been proposed. However, taken in isolation, these markers are limited in terms of arrhythmic risk stratification. This report analyses the value of the different parameters of ventricular repolarisation in the identification of high risk: QT dispersion, QT dynamics and T wave alternans. The dispersion of the QT interval is a marker of unhomogenous ventricular depolarisation. This concept must be applied differently in such pathologically dissimilar diseases such as myocardial infarction, cardiomyopathy or the long QT syndrome. Moreover, methodological problems make the interpretation of many experimental studies very delicate. Frequency dependence of the QT helps select high risk patients after myocardial infarction or with dilated cardiomyopathy. A common feature of pathological ventricular myocardium is the more pronounced frequency-dependency of the QT interval. The predictive value of this new index should be evaluated and compared with other non-invasive risk factors in prospective trials. Studies of T wave alternans in selected high risk populations, essentially patients with coronary artery disease and dilated cardiomyopathy, have shown this parameter to be predictive of arrhythmia. The predictive value requires confirmation in much larger populations at lower levels of risk of arrhythmia and sudden death in prospective trials. A new field of research has opened up in the study of ventricular repolarisation. Many studies have been undertaken on the duration of the QT interval, the morphology of the QT (including T wave alternans and post-pause changes) and, finally, the dynamics of the QT interval. By regrouping, analysing and using these data correctly, we should be able to identify new markers of high arrhythmic risk.