Sudden unexpected death in epilepsy (SUDEP) has been recognised for centuries. The precise frequency of occurrence is not well defined. Education of medical professionals is needed, so that death certificates and coronial inquests may appropriately, correctly and consistently record SUDEP as the case of death. Correct identification will then allow further investigation of this misunderstood, and often ignored, epilepsy complication. SUDEP incidence may be increasing, either as a result of increased recognition, or possibly due to a real increase in incidence. All currently available antiepileptic drugs (AEDs) have been associated with SUDEP, and current opinion assumes that the relative proportion of patients suffering SUDEP is representative of average AED usage type for a particular time and locality, however, recently analysed data suggest a strong bias towards carbamazepine. A review of Cardiff Epilepsy Unit data shows that carbamazepine was disproportionately represented in patients suffering SUDEP. In this series, 11 of the 14 SUDEP patients were taking carbamazepine at the time of death. This was calculated as 79% of all patients, compared to average carbamazepine usage by all other Cardiff Epilepsy Unit patients of 38%. The data also indicate that one patient was not taking any drug therapy, and died during his first seizure, reducing the number of evaluable 'drug usage' patients to 13, and increasing the proportion taking carbamazepine at the time of death to 85%, (P < 0.01). Possible mechanisms include carbamazepine induced lengthening of the ECG Q-T interval combined with a mild pro-arrhythmic effect of epileptic seizure discharges, and consequent transient cardiac instability leading to arrhythmic death. Or alternatively, excessive post-seizure brainstem inhibition might result in blunting or transient abolition of central hypoxic and hypercarbic respiratory drive, with consequent post-ictal respiratory arrest, subsequent exacerbation of hypoxia, further cardiac destabilisation and death due to hypoxia/failed re-establishment of respiration and terminal cardiac arrhythmia. Current knowledge about SUDEP remains poor. Education is needed so that case ascertainment can be correctly documented. Delineation of the precise mechanisms involved should lead to definitive prevention strategies. Evaluation of carbamazepine as a significant causative factor in SUDEP is also needed.