Knowledge of the recurrence risk following a first unprovoked seizure and the predictors of that risk are necessary for rational treatment decisions. Published estimates of recurrence risk range from 23% to 71%. In a meta-analysis of 16 reports, three methodologic factors explained much of the reported variation: (1) study inclusion criteria, ie, whether patients were enrolled at the time of their first seizure or if patients with prior seizures were included; (2) retrospective versus prospective ascertainment of patients; (3) the interval between the first seizure and the time at which risk was assessed. The average recurrence risk across the 16 studies was 51%. The risk was 40% and 52% in prospective and retrospective studies that employed first-seizure methods and 67% in non-first seizure studies. At or near 2 years following the first seizure, the recurrence risk was 36% and 47% in prospective and retrospective first-seizure studies. The distribution of prognostic factors was also important. Seizure etiology and the EEG were the strongest predictors of recurrence distinguishing between patient subgroups, with recurrence risks as low as 24% and as high as 65%. Partial seizures were associated with an increased recurrence risk, but not consistently. There is considerable agreement among studies concerning the recurrence risk following a first seizure, and much of the discrepancies among studies can be explained by differences in study methods and distributions of important prognostic factors.