The cholera toxin from Vibrio cholerae (CT) and the 80% homologous heat- labile toxin of Escherichia coli (LT) are two well-known cases of sugar- binding proteins. The GM1:toxin complexes were chosen as test cases for the elaboration of a computational approach to the modeling of protein-saccharide interactions. The reliability of the method was evaluated on the LT:lactose complex. A model of this complex was built by performing a MC/EM conformational search of the sugar moiety within the binding pocket of LT, using the AMBER force field and the GB/SA solvation model. The results are a reasonable reproduction of the reported X-ray structure of the complex. The same protocol was then applied to the LT:GM1 complex. The calculations were performed on a substructure that includes the protein shell within 5 Å from GM1, three water molecules solvating Glu-51 carboxylate, and two Water molecules at crystallographic sites 2 and 3. A satisfactory agreement was found with the recently published X-ray structure of the CT:GM1 complex. All the relevant interactions between the sugar and the residues involved in binding are well reproduced by the calculations. These results suggest that the substructure here identified can be taken as a realistic representation of the toxin binding surface and that the method presented in this paper can be used as a predictive tool in designing artificial LT (CT) binders and thus potential anticholera drugs.