Two methods are available for exploring arrhythmias in cardiac patients who are at risk of sudden death: Holter monitoring and invasive electrophysiology. Despite numerous studies, the predictive value of these techniques, in terms of prognosis, remains poor for many reasons. Neither technique considered individually can give reliable prognostic indications simply because each technique addresses different issues which are only partially involved in the mechanism of sudden death. Invasive electrophysiology, by artificially provoking an arrhythmia, detects the potential substrate which may ultimately lead to lethal arrhythmias. Although this is an important technique it is insufficient because merely identifying the substrate for an arrhythmia does not necessarily mean that arrhythmia will occur. On the other hand, ambulatory ECG allows monitoring of spontaneous arrhythmias which may be considered as potential initiating factors in arrhythmias. However, even if initiating factors and potential substrates are present, they are not sufficient conditions to cause lethal arrhythmias to occur. When there is an opportunity to scrutinize the mechanism of arrhythmias which are indeed lethal, as in sudden death, it appears that the lethal event results from the intervention of a new factor which was either absent or not considered during preceding investigations. In coronary patients, curiously, ischemia more often provokes cardiac arrest or an electromechanical dissociation rather than a ventricular tachycardia or fibrillation. Sudden death is not infrequently of iatrogenic origin, because of the arrhythmogenic effect of powerful antiarrhythmic drugs. More important, ventricular fibrillation often occurs in the setting of a progressively increased sympathetic tone, which explains either the particular seriousness of a previously known arrhythmia or the occurrence of an arrhythmia which was never before observed.(ABSTRACT TRUNCATED AT 250 WORDS)