Glycosynthase mutants obtained from Thermotogamaritima were able to catalyze the regioselective synthesis of aryl beta-d-Galp-(1-->3)-beta-d-Glcp and aryl beta-d-Glcp-(1-->3)-beta-d-Glcp in high yields (up to 90 %) using aryl beta-d-glucosides as acceptors. The need for an aglyconic aryl group was rationalized by molecular modeling calculations, which have emphasized a high stabilizing interaction of this group by stacking with W312 of the enzyme. Unfortunately, the deprotection of the aromatic group of the disaccharides was not possible without partial hydrolysis of the glycosidic bond. The replacement of aryl groups by benzyl ones could offer the opportunity to deprotect the anomeric position under very mild conditions. Assuming that benzyl acceptors could preserve the stabilizing stacking, benzyl beta-d-glucoside firstly assayed as acceptor resulted in both poor yields and poor regioselectivity. Thus, we decided to undertake molecular modeling calculations in order to design which suitable substituted benzyl acceptors could be used. This study resulted in the choice of 2-biphenylmethyl beta-d-glucopyranoside. This choice was validated experimentally, since the corresponding beta-(1-->3) disaccharide was obtained in good yields and with a high regioselectivity. At the same time, we have shown that phenyl 1-thio-beta-d-glucopyranoside was also an excellent substrate leading to similar results as those obtained with the O-phenyl analogue. The NBS deprotection of the S-phenyl group afforded the corresponding disaccharide quantitatively.