The National General Practice Study of Epilepsy is a prospective population-based cohort study of 1195 patients with newly diagnosed or suspected epileptic seizures. At the time of initial classification (6 months after notification), 104 patients were excluded. Of the remaining 1091 patients, 220 (20% [95% confidence interval 18-23%]) had febrile seizures, 564 (52% [49-55%]) definite epileptic seizures, and 228 (21% [19-23%]) possible epilepsy. In the definite epilepsy group the proportions of males and females were similar, 25% (21-28%) were younger than 15 years and 24% (21-28%) were 60 years or older. The definite seizures were classified as cryptogenic in 62% (58-66%), remote symptomatic in 21% (18-25%), and acute symptomatic in 15% (12-18%). The aetiology of epilepsy was vascular disease in 15% (12-18%) and tumour in 6% (4-8%). Among older subjects the proportion with an identifiable cause was much higher: 49% (41-58%) were due to vascular disease and 11% (6-16%) to tumour. Only 252 (45% [41-49%]) of the 564 patients with definite epileptic seizures were registered at the time of their first seizure. 52% (48-56%) of the patients had partial or secondarily generalised seizures, and only 39% (35-43%) seizures generalised from the outset.