Objective: To determine the frequency and types of significant, sustained arrhythmias in a mixed ICU. Design and setting: Prospective, observational study in a medical-cardiological-postoperative ICU at a university hospital. Patients: 133 consecutive patients with arrhythmias. Measurements and results: All patients had continuous ECG monitoring and automatic arrhythmia detection. We assessed: (a) sustained (>30 s) tachyarrhythmias; (b) all tachyarrhythmias requiring therapy; (c) bradycardias of fewer than 40 beats/min or requiring intervention. There were 310 arrhythmia episodes: 278 tachyarrhythmias (108 narrow-QRS complex, 168 wide-QRS complex; 179 regular, 97 irregular) and 32 bradycardias. Of the 278 tachycardias in 54 patients, 135 (48.6%) were ventricular. There were 13 episodes of torsade de pointes (4.67%) in five patients. Of the 278 tachycardiac episodes 83 were atrial fibrillation (29.8%, 63 patients), 10 atrial flutter (3.6%, 8 patients), 21 supraventricular tachycardias (7.55%, 7 patients), and 2 ectopic junctional tachycardia (0.72%, 1 patient). The number of patients showing significant arrhythmias was comparable over the years (11–12/1996: 4/28 [14.3], 1997: 52/302 [17.2%], 1998: 55/286 [19.2%], 22/140 [15.7%] 1–7/1999). The ICU stay was significantly longer in arrhythmia patients than in 623 patients without arrhythmias (median 4 vs. 14 days), and there was a trend towards higher mortality (40/133, 30.8%, vs. 132/623, 21.2%, P=0.061, log-rank). Conclusion: Only one-fifth of patients in this mixed ICU had significant arrhythmias, taking a contemporary definition of arrhythmias. Ventricular tachycardia and atrial fibrillation were the most frequent arrhythmias.