BACKGROUND: The quality of epilepsy care has often been noted to be poor and fragmentary. Most people with epilepsy are solely under the care of their general practitioner (GP). Many patients report medication side-effects and poor seizure control. Most GPs accept responsibility for epilepsy care; however, many report problems with knowledge of epilepsy and nearly all support guidance on epilepsy management. AIM: To determine whether a GP-completed prompt and reminder card is effective in improving the quality of epilepsy care when used opportunistically. DESIGN OF STUDY: Primary care-based pragmatic cluster-randomised controlled trial. SETTING: People with active epilepsy (n = 1275) from 82 practices. METHOD: Practices were randomly categorised as 'control', 'doctor-held card' (card in patient records), or 'patient-held card' practices. RESULTS: Compared with control practices, recording of seizure frequency was significantly increased in doctor-held card practices (57.4% versus 42.8%, P = 0.003) but not in patient-held card practices (44.6% versus 42.8%). No differences were found in the proportion of seizure-free patients (doctor-held card [56.0%] versus control [51.5%]; patient-held card [58.1%] versus control [51.5%]) or in the proportion on monotherapy. Patients in both intervention groups reported more medication-related side-effects and patients in doctor-held card practices were less satisfied with information provision about epilepsy. Participating GPs found the card useful. The doctor-held card was retrieved and completed more often than the patient-held card. CONCLUSIONS: A doctor-held prompt and reminder card is effective in improving the recording of key clinical information for people with epilepsy, is felt to be useful by GPs, and is completed more often than a patient-held card. However it does not improve outcomes and may result in less patient-centred care.