Members of the genus Mycobacterium cause devastating human diseases, including tuberculosis. Mycobacterium tuberculosis can resist some antibiotics because of its durable and impermeable cell envelope. This barrier is assembled from saccharide building blocks not found in mammals, including galactofuranose (Galf). Within the cell envelope, Galf residues are linked together to afford an essential polysaccharide, termed the galactan. The formation of this polymer is catalyzed by the glycosyltransferase GlfT2, a processive carbohydrate polymerase, which generates a sequence-specific polysaccharide with alternating regioisomeric beta(1-5) and beta(1-6) Galf linkages. GlfT2 exhibits high fidelity in linkage formation, as it will terminate polymerization rather than deviate from its linkage pattern. These findings suggest that GlfT2 would prefer an acceptor with a canonical alternating beta(1-5) and beta(1-6) Galf sequence. To test this hypothesis, we devised a synthetic route to assemble oligosaccharides with natural and non-natural sequences. GlfT2 could elongate each of these acceptors, even those with non-natural linkage patterns. These data indicate that the glycosyltransferase is surprisingly promiscuous in its substrate preferences. However, GlfT2 did favor some substrates: it preferentially acted on those in which the lipid-bearing Galf residue was connected to the sequence by a beta(1-6) glycosidic linkage. The finding that the relative positioning of the lipid and the non-reducing end of the acceptor influences substrate selectivity is consistent with a role for the lipid in acceptor binding. The data also suggest that the fidelity of GlfT2 for generating an alternating beta(1-5) and beta(1-6) pattern of Galf residues arises not from preferential substrate binding but during processive elongation. These observations suggest that inhibiting the action of GlfT2 will afford changes in cell wall structure.