Early investigators suggested that ventricular fibrillation without heart failure in acute myocardial infarction was reliably preceded by warning arrhythmias, and that suppression of such arrhythmias with intravenous lidocaine could avoid the need for resuscitation. While the efficacy and safety of lidocaine have been substantiated, the reliability of warning arrhythmias as predictors for primary ventricular fibrillation has not. We present data showing that the risk of primary ventricular fibrillation is most dependent on the patient's age and the interval since the onset of his symptoms, rather than on the presence of warning arrhythmias. We have estimated that lidocaine prophylaxis would have to be given to about 12 patients in the highest risk group (patients under age 50 and within six hours of the onset of symptoms), compared to about 400 patients in the lowest risk group (patients above age 70 and more than 24 hours since the onset of symptoms), to prevent one episode of primary ventricular fibrillation in each group. We propose that these risk stratifications, as adapted to the conditions in specific hospitals, provide the most rational approach to lidocaine prophylaxis of primary ventricular fibrillation.