In animal studies reperfusion of coronary arteries is commonly accompanied by ventricular arrhythmias. It is not certain, however, whether ventricular arrhythmias can be used as a reliable non-invasive marker of reperfusion in humans. Two-channel Holter recordings were obtained from the start of an intravenous infusion of streptokinase until coronary angiography (2.8 (2.7) hours (mean SD)) afterwards) in 57 patients with acute myocardial infarction of less than four hours who were generally not treated with antiarrhythmic drugs. Ventricular arrhythmias occurred in 21 (37%) of the 57 patients: accelerated idioventricular rhythm in 13 patients and non-sustained ventricular tachycardia in 15 patients. Seven patients had both accelerated idioventricular rhythm and non-sustained ventricular tachycardia. Coronary angiography showed a patent infarct-related vessel in 12 (92%) of the 13 patients with accelerated idioventricular rhythm (95% confidence interval 66 to 99%), in 22 (50%) of the 44 patients without accelerated idioventricular rhythm (95% CI 34 to 66%), in 11 (73%) of the 15 patients with non-sustained ventricular tachycardia (95% CI 45 to 92%), and in 23 (55%) (95% CI 39 to 71%) of the 42 patients who did not have non-sustained ventricular tachycardia. Seventeen (81%) of the 21 patients with accelerated idioventricular rhythm, or non-sustained ventricular tachycardia, or both, had a patent infarct-related vessel (95% CI 58 to 94%) as did 17 (47%) of the 36 patients with no ventricular arrhythmia (95% CI 29 to 65%). In patients with accelerated idioventricular rhythm after thrombolysis the infarct-related vessel is almost certain to be patent; but the infarct-related coronary artery can still be patent when no arrhythmia is seen.