Background To date, considerable controversy exists regarding noninvasive arrhythmia risk stratification in idiopathic dilated cardiomyopathy (IDC). Methods and Results Between 1992 and 1997, 202 patients with IDC without a history of sustained ventricular tachycardia (VT) underwent echocardiography, signal-averaged electrocardiogram (ECG), and 24-hour Holter ECG in the absence of antiarrhythmic drugs. During 32 +/- 15 months of prospective follow-up, major arrhythmic events, including sustained VT, ventricular fibrillation, or sudden death, occurred in 32 (16%) of 202 patients. After adjusting for baseline medical therapy and antiarrhythmic therapy during follow-up, multivariate Cox regression analysis identified a left ventricular (LV) end-diastolic diameter greater than or equal to 70 mm and nonsustained VT on Hotter as the only independent arrhythmia risk predictors. The combination of an LV end-diastolic diameter greater than or equal to 70 mm and nonsustained VT was associated with a 14.3-fold risk for future arrhythmic events (95% confidence interval 2.3-90). To further elucidate the prognostic value of LV election fraction, multivariate Cox analysis was repeated with election fraction forced to remain in the model. In the latter model, an election fraction less than or equal to 30% combined with nonsustained Vi on Holter was found to be a significant arrhythmia risk predictor with a relative risk of 14.6 (95% confidence interval 2.2-97). Conclusions The combination of an LV end-diastolic diameter greater than or equal to 70 mm and nonsustained VT on Hotter, and the combination of LV election fraction less than or equal to 30% and nonsustained VT on Holter, identify a subgroup of patients with IDC with a 14-fold risk for subsequent arrhythmic events. These findings have important implications for the design of future studies evaluating the role of prophylactic defibrillator therapy in patients with IDC.