Enzyme-linked immunosorbent assay (ELISA) represents a bioanalytical strategy frequently used for rapid screening of mycotoxin deoxynivalenol (DON) in cereals and derived products. Due to a considerable affinity of some anti-DON antibodies to structurally similar DON metabolites, such as DON-3-glucoside (DON-3-Glc) and 3-acetyl-DON (3-ADON), a significant overestimation of DON concentrations may occur. A validation study of six commercial DON-dedicated ELISA kits, namely Ridascreen DON, Ridascreen FAST, DON, DON EIA, AgraQuant DON Assay, Veratox 5/5, and Veratox HS was carried out on wheat, barley, and malt matrices. Performance characteristics of all tested ELISAs were determined using aqueous solutions of DON, DON-3-Glc, and 3-ADON analytical standards, further with extracts of artificially spiked blank cereals, and finally with matrix-matched standards of all three compounds. In the final phase, the accuracy of data was assessed through a comparison of DON concentrations determined by particular ELISAs and reference ultra-high-performance liquid chromatography-tandem mass spectrometry method. For this purpose, both quality control materials and a comprehensive set of naturally and artificially contaminated samples of wheat, barley, and malt were analyzed. High cross-reactivities were proved for both DON-3-Glc and 3-ADON in the majority of examined assays, and moreover, a considerable contribution of some matrix components to overestimation of DON results was confirmed.