We retrospectively evaluated the first 100 patients who underwent mapping-guided subendocardial resection (SER) at our hospital for drug-refractory sustained ventricular tachycardia caused by coronary artery disease. There were 91 survivors of surgery with 200 morphologically distinct types of ventricular tachycardia. Eighty-three patients (91%) were cured of ventricular tachycardia by SER alone (60 patients or 66%) or by SER in combination with antiarrhythmic drug therapy (23 patients or 25%) (mean follow-up, 28 +/- 19 months). There were four late sudden deaths and four patients continued to have rare episodes of spontaneous ventricular tachycardia after surgery despite receiving antiarrhythmic drugs. Factors associated with failure of SER alone to cure ventricular tachycardia were presence of disparate sites of ventricular tachycardia origin (greater than 5 cm between mapped sites of origin; 64% vs 30% failure rate) and presence of multiple morphologically distinct spontaneous tachycardias (47% vs 25% failure rate). A log-linear model of multivariate analysis identified disparate sites of origin of ventricular tachycardia and the absence of a discrete left ventricular aneurysm as the only independent variables associated with failure of surgery alone. Inferior wall site of origin (41% vs 12% failure) and right bundle branch block morphology of ventricular tachycardia (20% vs 7% failure) were also significantly associated with failure of surgery to cure ventricular tachycardia. Mapping-guided SER is a highly effective mode of treatment for drug-refractory ventricular tachycardia, despite the existence of subgroups of patients with higher-than-average surgical failure rates.