Using strictly defined criteria of significant arrhythmias, long term electrocardiographic recording has been evaluated for confirmation of arrhythmias as the cause of cerebral symptoms in 81 patients with suspected Adams-Stokes syndrome. Extension of long term electrocardiographic recording for more than 24 hours gives marginal additional information at a high cost. Among 43 patients monitored until symptoms appeared, non cardiogenic causes were confirmed in 20 of 22 patients because the recording showed normal rhythm during symptoms. Fifteen of 21 patients with a significant arrhythmia during an asymptomatic 24 hour recording later had the same arrhythmia during symptoms. Of 38 patients who failed to develop symptoms, 21 had a significant arrhythmia detectable within 24 hours and 23 when 48 hours of recording were analysed.