The challenge of preventing arrhythmic sudden death is one of the major issues in today's treatment of heart failure. To achieve this ambitious goal, an accurate selection of the candidates for sudden death is needed on the one hand, while on the other hand, the assessment of the real cost/benefit ratio of the implantable cardioverter-defibrillator in selected patients, as compared to ACE inhibitors, beta-blockers and antiarrhythmic drug therapy, should no longer be delayed. As is well known, the incidence of sudden death is higher in ischemic dilated cardiomyopathy than it is in non-ischemic dilated cardiomyopathy. Moreover, tachyarrhythmic sudden death is prevalent in NYHA classes I and II (80%), whereas its incidence is lower (50%) in NYHA classes III and IV, since bradyarrhythmia, electromechanical dissociation and thromboembolic events characterize the other 50% of sudden deaths in patients in the latter NYHA class. The stratification of arrhythmic risk in non-ischemic dilated cardiomyopathy is questionable from any point of view, considering the poor predictive power of invasive and non-invasive indexes. However, some subgroups of high-risk patients should be selected, such as patients waiting for heart transplant or those with a severe disease but without an extreme degree of ventricular dysfunction, in whom the prognosis in terms of pump failure events is better and life expectancy is longer if the risk of arrhythmia is properly assessed and sudden death prevented. Consequently, the ICD implant may be effective in order to pursue the aim of reducing the tachyarrhythmic and bradyarrhythmic mortality in patients with a more severe disease and of minimizing the tachyarrhythmic risk in those with a less severe disease. Further studies will be developed to identify the ideal candidates for ICD implants.