The objective of this study was to determine clinical and polysomnographic risk factors that might be early predictors for the development of postnatal epilepsy in a cohort of infants with seizures. The study sample included 158 infants who presented two or more clinically proven seizures. Gestational, perinatal, and polysomnographic data were obtained retrospectively. A questionnaire designed to detect patients with epilepsy in the community was prospectively given to all families, and the positive cases were reassessed for confirmation of epilepsy. Epilepsy rate after neonatal seizures was 22% within 12 months of follow-up and 33.8% within 48 months. Transient electrolytic imbalance and perinatal asphyxia were the most frequent etiologic factors associated with neonatal seizures. More than one seizure type was detected in 17.3% (n = 22) of cases and strongly associated with central nervous system infection (relative risk [RR] = 3.02, 95% confidence interval [CI] = 1.24-7.40, P = 0.02). Focal symptomatic epilepsy (P = 0.01) and syndromes not determined as focal or generalized (P = 0.04) were also associated with central nervous system infection. Abnormal polysomnographic recordings (P = 0.09) and abnormal neurologic examination on discharge (P < 0.01) were correlated with postnatal epilepsy. No differences were observed between premature and term infants concerning outcome. Neonatal seizures were associated with a high incidence of postnatal epilepsy in the cohort, including epileptic syndromes with catastrophic evolution. Abnormal neurologic examination on discharge was a good predictor of an unfavorable outcome and abnormal polysomnographic recording a moderate predictor.