The diagnosis of Lyme borreliosis is challenging because of the often non-specific symptoms and persisting antibodies after infection. We investigated the diagnostic characteristics of two enzyme-linked immunosorbent assays (ELISAs) and an immunoblot for the detection of Borrelia-specific serum antibodies using different test strategies in individuals with and without antibiotic treatment for Lyme borreliosis. This retrospective study included healthy individuals, patients with active Lyme neuroborreliosis and patients treated for Lyme neuroborreliosis. Two ELISAs were compared: the C6 ELISA and the SERION ELISA. Equivocal and positive results were confirmed by immunoblot. We included 174 healthy individuals, of whom 27 (15.5%) were treated for Lyme borreliosis in the past, 36 patients were treated for Lyme neuroborreliosis and 27 patients had active Lyme neuroborreliosis. All the active Lyme neuroborreliosis patients were reactive in both ELISAs (100% sensitivity); less reactivity was seen in the other three groups (range 17.7% to 69.4%). The concordance between the ELISA results was high in active Lyme neuroborreliosis patients (26/27; 96.3%) and healthy individuals (131/147; 89.1%), but lower in treated healthy individuals (18/27; 66.7%) and treated Lyme neuroborreliosis patients (18/36; 50.0%) (p ≤ 0.005). This study showed that antibiotic treatment against Lyme borreliosis was strongly associated with discordant ELISA and test strategy results (odds ratio: 10.52; p < 0.001 and 9.98; p = 0.014, respectively) suggesting antibiotic treatment influences the pace at which the various antibodies directed to the different antigens used in both ELISAs wane. Among treated neuroborreliosis patients, the SERION ELISA stayed positive for a longer period after infection compared to the C6 ELISA. This should be taken into consideration when requesting and/or interpreting Lyme serology.