Molecular modeling has been used to assist in the development of a novel series of potent glycogen phosphorylase inhibitors based on a phenyl diacid lead, compound 1. In the absence of suitable competitive binding assays, compound 1 was predicted to bind at the AMP allosteric site based on superposition onto known inhibitors which bind at different sites in the enzyme and analyses of the surrounding protein environment associated with these distinct sites. Possible docking modes of compound 1 at the AMP allosteric site were further explored using the crystal structure of rabbit muscle glycogen phosphorylase complexed with a Bayer diacid compound W1807 (PDB entry 3AMV). Compound 1 was predicted to interact with positively charged arginines at the AMP allosteric site in the docking model. Characterization of the binding pocket by a grid-based surface calculation of the docking model revealed a large unfilled hydrophobic region near the central phenyl ring, suggesting that compounds with larger hydrophobic groups in this region would improve binding. A series of naphthyl diacid compounds were designed and synthesized to access this hydrophobic cleft, and showed significantly improved potency.