The cell wall mycolyl-arabinogalactan-peptidoglycan complex is essential in mycobacterial species, such as Mycobacterium tuberculosis and is the target of several antitubercular drugs. For instance, ethambutol targets arabinogalactan biosynthesis through inhibition of the arabinofuranosyltransferases Mt-EmbA and Mt-EmbB. A bioinformatics approach identified putative integral membrane proteins, MSMEG2785 in Mycobacterium smegmatis, Rv2673 in Mycobacterium tuberculosis and NCgl1822 in Corynebacterium glutamicum, with 10 predicted transmembrane domains and a glycosyltransferase motif (DDX), features that are common to the GT-C superfamily of glycosyltransferases. Deletion of M. smegmatis MSMEG2785 resulted in altered growth and glycosyl linkage analysis revealed the absence of AG alpha(1-->3)-linked arabinofuranosyl (Araf) residues. Complementation of the M. smegmatis deletion mutant was fully restored to a wild-type phenotype by MSMEG2785 and Rv2673, and as a result, we have now termed this previously uncharacterized open reading frame, arabinofuranosyltransferase C (aftC). Enzyme assays using the sugar donor beta-d-arabinofuranosyl-1-monophosphoryl-decaprenol (DPA) and a newly synthesized linear alpha(1-->5)-linked Ara(5) neoglycolipid acceptor together with chemical identification of products formed, clearly identified AftC as a branching alpha(1-->3) arabinofuranosyltransferase. This newly discovered glycosyltransferase sheds further light on the complexities of Mycobacterium cell wall biosynthesis, such as in M. tuberculosis and related species and represents a potential new drug target.