Continuous electrocardiography during the first 24 hours of a stay in a coronary care unit was used to record ventricular arrhythmias during treatment with alteplase (recombinant tissue plasminogen activator) or placebo. Recordings were made on 378 of the 436 patients admitted to a double blind trial of alteplase or placebo in one participating centre of the Anglo-Scandinavian study of early thrombosis (ASSET), patients being selected according to the availability of recorders. Of these, 309 (158 given alteplase and 151 placebo) had greater than 5 hours of analysable data. Most of the arrhythmias were recorded in patients with an in hospital diagnosis of myocardial infarction. Ventricular couplets and ventricular tachycardia were significantly more common in the patients treated with alteplase. Further, in patients with myocardial infarction who had ventricular extrasystoles, couplets, or ventricular tachycardia type a, the number of hours in which each arrhythmia was recorded was significantly higher in the alteplase group. The various ventricular arrhythmias in the alteplase group tended to cluster in the first 4-12 hours of the recordings. During the first 24 hours admission there were four episodes of ventricular fibrillation in the alteplase group and five in the placebo group of taped patients. By one month there had been 18 deaths in these 309 patients (alteplase four, placebo 14). These bore no relation to any recorded arrhythmia. Clinical records for the patients with no or minimal tape data yielded six further episodes of ventricular fibrillation during the first 24 hours (three in the alteplase group and three in the placebo group). Of the total 436 patients, 10 of the 218 patients in the alteplase group had died by one month compared with 22 of the 218 patients treated with placebo. The use of alteplase increases the incident of non-life threatening ventricular arrhythmias. These results, however suggest that arrhythmia after thrombolysis in the pre-hospital phase may be less of a problem than it is perceived to be.