We prospectively studied the reliability and accuracy of the electroencephalogram as a predictor of the risk of recurrence within 2 years in 157 patients with untreated idiopathic first seizures. In all patients, a standard electroencephalogram and, if necessary, an electroencephalogram after partial sleep deprivation were obtained. All electroencephalograms were scored by one observer according to a fixed protocol. The finding of epileptic discharges was associated with a risk of recurrence of 83% (95% confidence interval, 69% to 97%) vs 41% (95% confidence interval, 29% to 53%) in patients with nonepileptic abnormalities and 12% (95% confidence interval, 3% to 21%) in patients in whom both electroencephalograms were normal. The sensitivity proved to be 48%. Interobserver agreement among four neurologists, who independently read 50 electroencephalograms, was found to be moderate. Predictive value for each observer, however, was good. We conclude that electroencephalogram findings may play a role in the decision to initiate or delay treatment after an idiopathic first seizure.