Although the nature of the blood groups A and B has been comprehensively studied for a long time, it is still unclear as to what exactly is the epitope that is recognized by antibodies having AB specificity, i.e. monoclonal and polyclonal antibodies which are capable of interacting equally well with the antigens GalNAcalpha 1-3(Fucalpha 1-2)Gal (A trisaccharide) and Galalpha 1-3(Fucalpha 1-2)Gal (B trisaccharide), but do not react with their common fragment Fucalpha 1-2Gal. We have supposed that besides Fucalpha 1-2Gal, A and B antigens have one more shared epitope. The trisaccharides A and B are practically identical from the conformational point of view, the only difference being situated at position 2 of Galalpha residue, i.e. trisaccharide A has a NHAc group, whereas trisaccharide B has a hydroxyl group (see formulas). We have hypothesized that the AB-epitope should be situated in the part of the molecule that is opposite to the NHAc group of GalNAc residue. In order to test this hypothesis we have synthesized a polymeric conjugate in such a way that de-N-acetylated A-trisaccharide is attached to a polymer via the nitrogen in position C-2 of the galactosamine residue. In this conjugate the supposed AB-epitope should be maximally accessible for antibodies from the solution, whereas the discrimination site of antigens A and B by the antibodies should be maximally hidden due to the close proximity of the polymer. Interaction with several anti-AB monoclonal antibodies revealed that a part of them really interacted with the synthetic AB-glycotope, thus confirming our hypothesis. Moreover, similar antibodies were revealed in the blood of healthy blood group 0 donors. Analysis of spatial models was performed in addition to identify the hydroxyl groups of Fuc, Galalpha, and Galbeta residues, which are particularly involved in the composition of the AB-glycotope.