The contamination of inactivated vaccine with non-structural proteins (NSPs) leads to a high false-positive rate, which is a substantial barrier to accurately differentiate foot-and-mouth disease virus (FMDV)-infected animals from vaccinated animals. To address this problem, a new chemiluminescence immunoassay (CLIA) method was developed to detect antibodies targeting the two recombinant epitope-based proteins located in 3A and 3B. The 3Aepitp-3Bepitp CLIA exhibited a diagnostic sensitivity of 94.0% and a diagnostic specificity of 97.5% for the detection of serum samples (naïve bovines, n = 52, vaccinated bovines, n = 422, infected bovines, n = 116) from animals with known status. The CLIA method also had a concordance rate of 88.1% with the PrioCHECK FMDV NSP ELISA based on the detection of 270 serum samples from the field. Importantly, the 3Aepitp-3Bepitp CLIA produced no false-positives when used to detect FMDV in samples from bovines that had been vaccinated up to five times, and it was demonstrated a low false-positive rate when the bovines had been vaccinated up to ten (2.15%) and fifteen times (5.93%). Therefore, the 3Aepitp-3Bepitp CLIA detects FMDV in samples from frequently vaccinated bovines with high accuracy and represents an alternative method to differentiate FMDV-infected and vaccinated bovines.