Each approach to the prevention of sudden death after myocardial infarction should be analysed not only according to its value in the risk stratification of patients but also for its potential impact in guiding therapy. With the results of 2 or more noninvasive tests [assessment of left ventricular ejection fraction (LVEF) by radionuclide ventriculography, grade of ventricular arrhythmias and heart rate variability by Holter monitoring, and ventricular activation by signal averaged electrocardiography], most patients may be stratified into 'very high' and 'very low' risk groups. For patients with 2 or more abnormal noninvasive tests, electrophysiological studies (EPS) may be recommended: if sustained ventricular tachycardia (VT) is not induced, patients may be reassured and left without antiarrhythmic therapy other than beta-blockers. For patients with low LVEF, treatment with beta-blockers may be recommended based on post hoc analysis of large prospective trials, while some of the randomised studies with angiotensin converting enzyme (ACE) inhibitors suggest that these agents may also reduce the risk of sudden death. For patients with high grade ventricular arrhythmias, beta-blockers and amiodarone may be recommended: the first, based on post hoc analysis of the Beta Heart Attack Trial, while data supporting the use of amiodarone come from prospective, yet small randomised studies. Empirical or Holter-guided therapy with class 1 antiarrhythmic drugs have not been found useful and may indeed be detrimental. For patients with inducible sustained VT, treatment should be guided by repeated EPS, as empirical antiarrhythmic therapy has not been found useful. However, the value of EPS-guided therapy remains to be proven. Patients with inducible sustained VT refractory to antiarrhythmic drugs are at very high risk. Implantation of an automatic defibrillator is an attractive option for these patients, to be confirmed by ongoing trials.