408 adults (age 16 and over) were followed up after their initial seizure. The actuarial risk of recurrence was highest in the early weeks. For those seen within the first week, the risk of recurrence was 52% by the end of 3 years. In univariate analysis, the only clinical variable that was associated with recurrence was the time of day at which the initial seizure occurred. There was a tendency, that did not reach statistical significance, for younger age (less than 50 years) and a family history of seizures of any type, including febrile convulsions, to be associated with recurrence. Such risk factors appear to be additive, so that the risk of recurrence at 1 year for a subject with all three factors is about three times that of a subject with none of them. Sex, type of seizure, and features of the electroencephalogram were not of predictive value. Computerised tomographic scanning revealed tumours in 3% of subjects, and these individuals were particularly likely to have recurrent seizures.