The aim of the study was to evaluate the sensitivity, specificity and the predictive values of repeated serum prolactin measurements in relation to epileptic seizures versus pseudoseizures. The method used was prospective measuring of serum prolactin from blood samples drawn (1) 15 min after seizure and (2) 2 hr after the first sample. Two parameters were used: the absolute maximal level; and the relative rise in blood level. In the study 38 had epilepsy (simple or complex partial seizures with or without secondary generalisation); and 20 had pseudo-epileptic seizures. In all cases, the diagnoses were made independent of the prolactin levels. In 30/38 (79%) of epilepsy patients and 17/20 (85%) of pseudoseizure patients, the diagnoses were corroborated by intensive EEG monitoring (video or cassette telemetry). There was a statistically significant rise in prolactin levels in both groups (p < 0.0001 and < 0.02, respectively), and also a significant difference between the two groups. However, repeated measurements in a number of patients (epilepsy: mean 1.5 measurements; pseudo; mean 2.1) showed also considerable intra-patient variations. The sensitivity for the maximal rise in pseudoseizures (5.5x) was only 20% and the negative predictive value 40%. For the cut-off in absolute level (1025 microU/ml), the corresponding figures were 34% and 44%, respectively. The rather limited discriminative power of prolactin measurements makes it of questionable value in discerning between epileptic and pseudo-epileptic seizures.