We assessed the frequency, causes and consequences of erroneous diagnosis of epilepsy, and the outcome of patients referred with 'refractory epilepsy', by retrospective analysis of the case records of 324 patients. The sample was divided into those exposed to anti-epileptic drugs (n = 184), of whom 92 were said to have refractory seizures, and those who had not received treatment (n = 140). The latter group is reported elsewhere. The overall misdiagnosis rate was 26.1% (46/184), with incomplete history-taking and misinterpretation of the EEG equally responsible. Side-effects were reported by 19/40, while unnecessary driving restrictions and employment difficulties were encountered by 12/33 and 5/33, respectively. Of those labelled 'refractory epilepsy', 12 did not have epilepsy. Sixteen were rendered seizure-free and 25 significantly improved by the optimal use of anti-epileptic drugs or surgery. Diagnostic and management services for patients with suspected and established epilepsy are suboptimal, with psychological and socio-economic consequences for individual patients. The resulting economic burden on the health and welfare services is probably substantial.