Background Patients with epilepsy are at risk of sudden unexpected death. Neu rogenic cardiac arrhythmias have been postulated as a cause. Electrocardiograms (ECG) can be monitored by use of an implantable loop recorder for up to 18 month s. We aimed to determine the frequency of cardiac arrhythmias in patients with r efractory focal seizures over an extended period. Methods 20 patients received a n implantable loop recorder at one hospital in the UK. Devices were programmed t o record automatically if bradycardia ( < 40 beats per min) or tachycardia ( > 1 40 beats per min) were detected. Additionally, in the event of a seizure, patien ts and relatives could initiate ECG recording with an external activator device. Data were analysed at regular intervals and correlated with seizure diaries. Fi ndings More than 220 000 patient hours were monitored over 24 months, during w hich ECGs were captured on implantable loop recorders in 377 seizures. One patie nt withdrew from the study. In 16 patients, median heart rate during habitual se izures exceeded 100 beats per min. Ictal bradycardia ( < 40 beats per min) was r are, occurring in eight (2 1% ) recorded events, in seven patients. Four pati ents (21% ) had bradycardia or periods of asystole with subsequent permanent pa cemaker insertion. Three of these four (16% of total) had potentially fatal as ystole. Interpretation Clinical characteristics of patients with peri ictal ca rdiac abnormalities are closely similar to those at greatest risk of sudden unex pected death in epilepsy. Asystole might underlie many of these deaths, which wo uld have important implications for the investigation of similar patients and af fect present cardiac pacing policies. Background Patients with epilepsy are at risk of sudden unexpected death. Neu rogenic cardiac arrhythmias have been postulated as a cause. Electrocardiograms (ECG) can be monitored by use of an implantable loop recorder for up to 18 months s. We aimed to determine the frequency of cardiac arrhythmias in patients with r efractory focal seizures over an extended period. Methods 20 patients received an implantable loop recorder at one hospital in the UK. Devices were programmed to record automatically if bradycardia (<40 beats per min) or tachycardia (> 1 40, beats per min) were detected. Additionally, in the event of a seizure, patien ts and relatives can ECG were recorded with an external activator device. Data were analyzed at regular intervals and correlated with seizure diaries. Fi ndings More than 220 000 patient ━ hours were monitored over 24 months, during w hich ECGs were captured on implantable loop recorders in 377 seizures. One patie nt withdrew from the study. In 1 Six patients, median heart rate during habitual seizures exceeded 100 beats per min. Ictal bradycardia (<40 beats per min) was r are, occurring in eight (2.1%) recorded events, in seven patients. Four pati ents (21 % of) had bradycardia or periods of asystole with subsequent permanent pa cemaker insertion. Three of these four (16% of total) had potentially fatal as ystole. Interpretation Clinical characteristics of patients with peri-ictal ca rdiac abnormalities are closely similar to those at greatest risk of sudden unex pected death in epilepsy. Asystole might underlie many of these deaths, which wo uld have important implications for the investigation of similar patients and af fect present cardiac pacing policies.