Glycosyltransferases (GTs) are responsible for the biosynthesis of glycans, the most abundant organic molecules in nature. Their biological relevance makes necessary the knowledge of their catalytic mechanism, which in the case of retaining GTs is still a matter of debate. After the initial proposal of a double-displacement mechanism with formation of a covalent glycosyl-enzyme intermediate (CGE), new experimental and computational data are pointing out to a front-side attack as a plausible alternative. The question is then why family GT6 members, like bovine α1,3-galactosyltransferase (α1,3-GalT), have a nucleophilic residue (Glu317) situated close to the anomeric carbon. To answer this and other questions, QM(DFT)/MM calculations on the entire α1,3-GalT:substrates system (and for the E317A/E317Q mutants) have been carried out. We describe a substrate-assisted mechanism for retaining GTs consisting of the stabilization of the developing negative charge on the β-phosphate by the hydrogen of the attacking hydroxyl group of the acceptor molecule. This interaction is impaired in the α1,3-GalT reactants, which explains why Glu317 is required to nucleophilically assist initial catalysis by "pushing" leaving-group departure. The presence of Glu317 opens the door to the possibility of a double-displacement mechanism in GT6 family. Our results suggest that in α1,3-GalT the substrate-assisted catalysis would be necessary in both mechanisms (for which we predict similar reaction rates), because the nucleophilic strength of Glu317 is reduced by the interactions it makes to ensure proper acceptor binding. Interestingly, the same effect would be found in the absence of the acceptor when Glu317 interacts with water molecules, which could explain the difficulties for isolating the CGE experimentally, and could be a strategy to avoid undesired hydrolysis of the donor substrate.