BACKGROUND Fragmented QRS (fQRS) on a 12-lead electrocardiogram (ECG) is associated with myocardial scar in patients with coronary artery disease (CAD). OBJECTIVE We postulated that fQRS is a predictor of cardiac events and mortality in patients who have known CAD or who are being evaluated for CAD. METHODS The cardiac events (myocardial infarction, need for revascularization, or cardiac death) and all-cause mortality were retrospectively reviewed in 998 patients (mean age 65.5 +/- 11.9 years, male 967) who underwent nuclear stress test. The fQRS on a 12-lead ECG included various RSR' patterns (> or =1 R' prime or notching of S wave or R wave) without typical bundle branch block in 2 contiguous leads corresponding to a major coronary artery territory. RESULTS All-cause mortality (93 [34.1%] vs 188 [25.9%]) and cardiac event rate (135 [49.5%] vs 200 [27.6%]) were higher in the fQRS group compared with the non-fQRS group during a mean follow-up of 57 +/- 23 months. A Kaplan-Meier survival analysis revealed significantly lower event-free survival for cardiac events (P <.001) and all-cause mortality (P = .02). Multivariate Cox regression analysis revealed that significant fQRS was an independent significant predictor for cardiac events but not for all-cause mortality. The Kaplan-Meier survival analysis showed no significant difference between fQRS and Q waves groups for cardiac events (P = .48) and all-cause mortality (P = .08). CONCLUSIONS The fQRS is an independent predictor of cardiac events in patients with CAD. It is associated with significantly lower event-free survival for a cardiac event on long-term follow-up.