The predictivity of raised plasma prolactin (PRL) concentrations in differentiating seizure from syncopal attack was prospectively assessed in all patients consecutively admitted to the Clinica Neurologica of Brescia, Italy in a 12-month period who fulfilled the criteria for either a seizure or syncopal attack. Postictal plasma prolactin concentration (P1) was assessed as soon as possible after the event. Three further assessments were performed: P2 was sampled 1 h after P1, P3, and P4 were sampled in the morning for the next 2 days. Patients who had had a seizure showed significantly increased P1 concentrations, when P1 was sampled within 60 min of the attack. In seizure patients assessed > 1 h after the event, P1 was not significantly different from either P2, P3, or P4. In patients who had had a syncopal attack, PRL concentration never increased. In patients assessed < or = 60 min after the seizure, cutoff criterion of P1 exceeding by +3 SD the mean calculated on P2, P3, and P4 yielded a positive predictive value of 89% and a negative predictive value of 61%. These findings confirm that plasma prolactin concentration is highly predictive of true epilepsy but barely predictive of pseudoseizures.