Background In retrospective studies of sudden cardiac death survivors, the implantable cardioverter-defibrillator (ICD) compares favorably with medical and surgical therapy. Thus, use of the conventional strategy of starting treatment with antiarrhythmic drugs (AD), at least in certain patient categories, may be questionable. The goal of this study was to analyze the effectiveness of ICD implantation as first-choice therapy versus the conventional therapeutic strategy of starting with AD. Methods and Results Sixty consecutive survivors of cardiac arrest caused by old myocardial infarction were randomly assigned early ICD implantation (n=29) or conventional therapy (n=31). Baseline characteristics were similar in the two groups. Therapy in each patient was always guided by ECG monitoring, exercise testing, and programmed electrical stimulation (PES). Primary end points (main outcome events, including death, recurrent cardiac arrest, and cardiac transplantation), number of invasive procedures and antiarrhythmic therapy changes, and duration of hospitalization were compared. Median follow-up was 24 months (mean, 27 months). In the early ICD group, 4 patients (14%) died, all of cardiac causes. In the conventional group, 20 patients failed AD and subsequently underwent map-guided ventricular tachycardia (VT) surgery (6 patients) or ICD implantation (14 patients). Of the 6 VT surgery patients, 1 died, 1 had cardiac transplantation, and 1 had an ICD implantation because of persistent inducibility despite the addition of AD. Of the 11 patients who remained on AD as sole therapy, 2 died in the hospital before they could be retested by PES, leaving 9, judged adequately protected by AD alone. Of those, 5 died, and 1 survived recurrent cardiac arrest followed by ICD implantation. In total, 16 conventionally treated patients ended up with late ICD implantation, 3 of whom died. Thus, total mortality in the conventional group was 11 patients (35%): 4 died suddenly, 5 died of heart failure, and 2 died of noncardiac causes. Comparison of the main outcome events in both strategies showed a significant difference in favor of early ICD implantation (hazard ratio, 0.27; 95% CI, 0.09 to 0.85; P=.02). In addition, the early ICD group underwent fewer invasive procedures (median, 1 versus 3; P<.0001), had less therapy changes (P<.0001), and spent fewer days in hospital (median, 34 versus 49; P=.02). Conclusions These data suggest that ICD implantation as first choice is preferable to the conventional approach in survivors of cardiac arrest caused by old myocardial infarction. Conventionally treated patients are likely to end up with an ICD, and those who remain on AD as sole therapy have a high risk of death regardless of efficacy assessment, including PES. .