The long-term survival data in patients with coronary artery disease and a history of malignant ventricular arrhythmia, defined as noninfarction ventricular fibrillation (VF) or hemodynamically compromising ventricular tachycardia (VT) followed for up to 9 years, were analyzed. In this group of 161 patients there was a total of 57 deaths, of which 35 (63%) were sudden. Life-table analysis demonstrated a 10% sudden death rate for all patients in the first year and a 7% annual rate in the subsequent 4 years. In patients managed noninvasively, the overall mortality rate was 27% over 9 years, or 3% per year. Suppression of ventricular tachycardia on both ambulatory monitoring and exercise testing was associated with improved survival. In patients evaluated by electrophysiologic testing the sudden death rate was 1.4% per year over an average of 5 years. This survival rate was not different compared with the noninvasive group (p = 0.09). Measures of left ventricular dysfunction and the frequency of ventricular arrhythmia before and after drug therapy were associated with a risk of sudden cardiac death by univariate analysis. Multivariate regression analysis identified 4 variables as independent predictors of sudden cardiac death: rales (p = 0.009), the number of runs of VT during exercise testing while receiving antiarrhythmic drug therapy (p = 0.0003), a history of congestive heart failure (p = 0.0009) and the number of premature beats on Holter monitoring (p = 0.01). These findings support the concept that suppression of repetitive arrhythmia on Holter monitor and exercise testing is a marker for improved survival among patients with malignant ventricular arrhythmia. However, in the coronary population at high risk of sudden cardiac death, ventricular function is the most powerful predictor of survival and, when severely depressed, the control of ventricular arrhythmia does not significantly improve outcome.