Transcriptional regulatory proteins engineered to respond to specific stimuli find applications in genetic circuit design and metabolic engineering and enable customized genetic selections. In this study a systematic computational framework has been developed to study and engineer the effector binding specificity of the Escherichia coli transcriptional regulatory protein AraC. Simulation and optimization methods have been employed to accurately reflect the relative strengths with which wild-type AraC binds various compounds, and are being used to predict mutagenesis strategies resulting in altered binding selectivity. Our initial goal is to design proteins capable of distinguishing between D- and L-arabinose. Our model identifies mutations near the binding pocket that are likely to improve selectivity for the desired ligand. We compare binding scenarios with and without inclusion of water molecules in the binding pocket and find compensatory mutations with regard to size and polarity. Experimentally, these positions are targeted for mutagenesis to multiple or all possible amino acids, resulting in large mutant libraries. In E. coli, a selection system has been developed to facilitate the identification of AraC variants which activate transcription in the presence of the desired effector molecule. A negative selection is then employed to eliminate constitutive mutants or other false positives, as well as to enrich specificity by identifying mutants with reduced sensitivity to similar, “competing” compounds (e.g., L-arabinose).