Like most enzymes catalyzing the first step in sugar metabolism, the broad substrate acceptance of XR is evolutionarily advantageous to organisms, but disadvantageous for a chemical process attempting to minimize byproduct formation. The goal of this project is therefore to reverse the evolutionary advantage conferred by using protein engineering techniques to decrease the specificity of a XR toward L-arabinose, while maintaining high activity toward D-xylose. The Neurospora crassa XR (NcXR) was chosen for engineering work due to several favorable properties over other XRs, in addition to its innate >2-fold catalytic efficiency toward D-xylose than L-arabinose. A directed evolution strategy was developed that consists of a combined structure-function based semi-rational design involving active site residue mutagenesis followed by random mutagenesis and selection for desired substrate specificity. After the first round of evolution, a mutant was identified with fourteen-fold preference for D-xylose over L-arabinose. To our knowledge this is the most xylose-specific XR identified or engineered to date. Further engineering rounds are currently underway on this template to further reduce its promiscuity.