Background: This is a retrospective study to determine the usefulness of serum prolactin levels in the epilepsy monitoring unit (EMU) for distinguishing psychogenic nonepileptic seizures (PNES) from epileptic seizures (ES). Methods: All the records of consecutive patients admitted to the EMU between 2008 and 2012 were reviewed. Patients with a diagnosis of PNES and ES were selected. Serum prolactin level was obtained within 20 minutes for all patients, and an elevated prolactin level was defined as twice the patient's baseline value. A total of 607 records were reviewed, and 389 patients were excluded because of incomplete information or lack of a clinical event. Results: A total of 218 patients were included in the analysis. A further 18 patients were excluded because of simple partial seizures or because the documented time of obtaining serum prolactin was not clear. A total of 146 patients had PNES, of which 42 had elevated prolactin levels, representing a 28.8% false-positive rate. All 22 patients with generalized tonic-clonic seizures had elevated serum prolactin; however, 5 of 32 patients with complex partial seizures did not have an elevated prolactin level, representing a 15.6% false-negative rate. Conclusions: Serum prolactin levels do not provide any additional support for distinguishing PNES from ES. Furthermore, serum prolactin levels in this setting add unnecessary blood draws and financial burden for the patients. Therefore, prolactin levels should not be obtained routinely in the EMU. Classification of evidence: This study provides Class III evidence that postevent elevation of serum prolactin occurs in 84.4% of patients with ES and 28.8% of patients with PNES.