OBJECTIVE: To report the prevalence of anti-neuronal antibodies in a prospective whole-nation cohort of children presenting with seizures before their third birthday. METHODS: This was a prospective population-based national cohort study involving all children presenting with new-onset epilepsy or complex febrile seizures before their third birthday over a 3-year period. Patients with previously identified structural, metabolic, or infectious cause for seizures were excluded. Serum samples were obtained at first presentation and tested for 7 neuronal antibodies using live cell-based assays. Clinical data were collected with structured proformas at recruitment and 24 months after presentation. In addition, patients with seizures and clinically suspected autoimmune encephalitis were independently identified by a review of the case records of all children <3 years of age in Scotland who had undergone EEG. RESULTS: Two hundred ninety-eight patients were identified and recruited and underwent autoantibody testing. Antibody positivity was identified in 18 of 298 (6.0%). The antibodies identified were GABA receptor B (n = 8, 2.7%), contactin-associated protein 2 (n = 4, 1.3%), glycine receptor (n = 3, 1.0%), leucine-rich glioma inactivated 1 (n = 2, 0.7%), NMDA receptor (n = 1, 0.3%), and GABA receptor A (n = 1, 0.3%). None of these patients had a clinical picture of autoimmune encephalitis. Seizure classification and clinical phenotype did not correlate with antibody positivity. CONCLUSIONS: Autoimmune encephalitis is very rare in early childhood. However serum neuronal antibodies are identified in 6.4% of children presenting with seizures at <3 years of age. Antibody testing should not be a routine clinical test in early childhood-onset epilepsy because, in the absence of other features of autoimmune encephalitis, antibody positivity is of doubtful clinical significance. Antibody testing should be reserved for patients with additional features of encephalitis.