A post hoc analysis of the Multicenter Automatic Defibrillator Implantation Trial (MADIT) II showed that patients with high mortality risk score did not benefit from implantable cardioverter-defibrillator (ICD) therapy. We sought to determine whether MADIT II risk score can identify patients with greater mortality in a nontrial “real-world” setting. We included 382 consecutive patients who received ICD for primary prevention of sudden cardiac death at the Minneapolis Veterans Affairs Medical Center from 2006 to 2010. MADIT II score was calculated by assigning 1 point each for age >70 years, New York Heart Association class >II, atrial fibrillation, QRS >0.12 seconds, and blood urea nitrogen level >26 mg/dl. Scores 0, 1 to 2, and ≥3 were classified as low, intermediate, and high risk, respectively. Of the 382 patients, 14% were low risk, 54% intermediate risk, and 32% high risk. After 3.0 ± 1.6 years of follow-up, incidence of appropriate ICD shocks was similar (p = 0.21) across MADIT II risk score categories. However, mortality rate was 21, 54, and 134 per 1,000 patient-year follow-up in low, intermediate, and high-risk patients, respectively, p <0.0001. Compared with low-risk patients, mortality was 6.4× greater in high-risk patients (hazard ratio 6.36, 95% confidence interval 1.9 to 20.5; p = 0.002). The c-index for the MADIT II score for predicting death was 0.69 (95% confidence interval 0.63 to 0.75). In conclusion, MADIT II risk score successfully identified patients with greater mortality in a nontrial-based, primary-prevention ICD cohort.