We evaluated the effect of caffeine on ventricular ectopic activity in a group of 50 consecutive patients with malignant ventricular arrhythmia. The clinical arrhythmia in these patients (mean age, 61 years) was recurrent ventricular tachycardia in 21 (42%), ventricular fibrillation in three (6%), and symptomatic nonsustained ventricular tachycardia in 26 (52%). Forty-two (84%) had either ischemic heart disease or cardiomyopathy. Each patient underwent two short-term drug trials on successive days, receiving either decaffeinated coffee mixed with 200 mg of caffeine or the decaffeinated drink alone. Continuous electrocardiographic recordings were made during the 30-minute control period, the three-hour observation period, and the hourly bicycle exercise tests. Forty-five patients (90%) exhibited ventricular couplets and 29 patients (58%) had salvos of ventricular tachycardia during the testing. However, no differences between the caffeine and decaffeinated trials were observed in either individual or group data on total or repetitive ventricular arrhythmia. Serum catecholamine levels reflected the average increase in serum caffeine level but were not associated with enhanced arrhythmia. We found no evidence that a modest dose of caffeine is arrhythmogenic, even among patients with known life-threatening arrhythmia.