Background: Improving the hydrolytic performance of hemicellulases on lignocellulosic biomass is of considerable importance for second-generation biorefining. To address this problem, and also to gain greater understanding of structure-function relationships, especially related to xylanase action on complex biomass, we have implemented a combinatorial strategy to engineer the GH11 xylanase from Thermobacillus xylanilyticus (Tx-Xyn). Results: Following in vitro enzyme evolution and screening on wheat straw, nine best-performing clones were identified, which display mutations at positions 3, 6, 27 and 111. All of these mutants showed increased hydrolytic activity on wheat straw, and solubilized arabinoxylans that were not modified by the parental enzyme. The most active mutants, S27T and Y111T, increased the solubilization of arabinoxylans from depleted wheat straw 2.3-fold and 2.1-fold, respectively, in comparison to the wild-type enzyme. In addition, five mutants, S27T, Y111H, Y111S, Y111T and S27T-Y111H increased total hemicellulose conversion of intact wheat straw from 16.7%(tot. xyl) (wild-type Tx-Xyn) to 18.6% to 20.4%(tot. xyl). Also, all five mutant enzymes exhibited a better ability to act in synergy with a cellulase cocktail (Accellerase 1500), thus procuring increases in overall wheat straw hydrolysis. Conclusions: Analysis of the results allows us to hypothesize that the increased hydrolytic ability of the mutants is linked to (i) improved ligand binding in a putative secondary binding site, (ii) the diminution of surface hydrophobicity, and/or (iii) the modification of thumb flexibility, induced by mutations at position 111. Nevertheless, the relatively modest improvements that were observed also underline the fact that enzyme engineering alone cannot overcome the limits imposed by the complex organization of the plant cell wall and the lignin barrier.