Carbohydrate binding modules (CBMs) play important biological roles in targeting appended catalytic modules to their dedicated substrate(s) within complex macromolecular structures such as the plant cell wall. Because of the large potential in ligand diversity within nature and our continually expanding knowledge of sequence-based information of carbohydrate-modifying enzymes, empirical determination of CBM binding specificity and identification of novel mechanisms in carbohydrate recognition by these proteins have become time-consuming and complicated processes. To help overcome these experimental hurdles, we present here a predictive model for family 6 CBMs (CBM6) that is based upon several factors, including phylogenetic relatedness, and structural and functional evidence. This analysis has determined that five regions within the binding site, termed A-E, play key roles in ligand selection and affinity. Regions A-C are located in a primary subsite and contribute mainly to binding energy and selection for O2, O3, and O4 equatorial hydroxyls. Region D appears to determine whether the CBM will interact with internal or terminal structures of the carbohydrate ligand. Region E displays the largest degree of variation and is thus predicted to make the most significant contribution to specificity. This model is supported by the biochemical properties and structure of a CBM6 from Clostridium cellulolyticum (CcCBM6), which we also report here. The protein bound specifically to xylose and the nonreducing of end of polymers containing this pentose sugar. The crystal structure of CcCBM6 in complex with xylose showed that a tyrosine residue made hydrophobic contacts with the unsubstituted C5 atom of xylose and sterically hindered decorations at this sugar ring position. The mechanism, by which the CBM recognizes xylose but not glucose, a specificity not previously observed in this family, supports our predictive model that holds that variation in region E plays a key role in the diverse ligand selection evident in CBM6.