No indisputable criteria for selecting candidates for direct surgery in postinfarction ventricular tachycardia are yet available. We prospectively tested the result of surgery in patients showing normokinesia or slight hypokinesia in at least 3 of 9 segments of the left ventricle as determined on two oblique projections of the left ventricular angiogram. Sixty consecutive patients with recurrent symptomatic ventricular tachycardia were stratified on the basis of this criterion; 50 candidates were thus eligible for surgery. Direct surgery (endocardial resection and/or cryoablation) was performed in 23 patients (after a mean of 4.1 antiarrhythmic drug trials), while 27 patients were maintained on drug treatment (at discharge, a mean of 3.7 drug trials). Surgery was map-guided in 19 patients and non-map-guided in four emergency cases. No patients died in the immediate postoperative phase, but two (9%) died during follow-up. The actuarial survival at 12 months was 93%; the surgery-alone cure rate for ventricular tachycardia was 78%. Of the 27 non-operated surgical candidates, two died during in-hospital drug testing and five after discharge (total 26%). Actuarial survival at 12 months was 78% and did not differ significantly from that of the operated patients. After discharge, 80% of the non-operated surgical candidates still using the drug at discharge remained free of recurrence. In the 10 drug-treated non-surgical candidates (at discharge, a mean of 3.8 drug trials), survival and the number of arrhythmia-free patients were markedly lower than in surgical candidates. This study shows that the segmental wall motion score is of great clinical value and can be applied for stratification of patients with ventricular tachycardia after myocardial infarction for either surgery or medical therapy.