Glycogen Phosphorylase is inhibited by a family of related compounds [purines, purine nucleosides, nucleotides, and certain heterocyclic compounds, e.g., flavin mononucleotide (FMN)] which bind to an allosteric site located at the surface of the enzyme 10 Å from the catalytic cleft at which glucose and glucose 1-phosphate are bound. The interaction of several such inhibitors, adenine, caffeine, adenosine, inosine, ATP, and FMN, with rabbit muscle Phosphorylase a in the glucose-inhibited form has been examined by X-ray crystallographic (difference Fourier) analysis at 3.0-and 2.5-Å resolution. The dissociation constant (Kd) for all of these ligands was determined by kinetic analysis and, for FMN, by fluorometry. The ΔS° and ΔH for association of FMN to Phosphorylase were derived from analysis of calorimetric data. We have synthesized the structural and thermodynamic data in order to arrive at a description of the energetics of the binding interaction and its specificity. The inhibitors associate with Phosphorylase by forming an intercalative complex in which the heterocyclic ring system is stacked between the aromatic side chains of F285 and Y612. When they do so, the inhibitor stabilizes the same conformation of residues 282–286 which binds α-d-glucose, an inhibitor which demonstrates synergism with the purine ligands. No other significant hydrogen-bonded contacts are made with the enzyme, and any polar or charged groups of the ligand (ribose, ribose phosphate, or ribityl phosphate) are solvated at the protein surface. There does not appear to be a single particularly favored orientation of the heterocyclic ring dipole within its binding pocket nor is there a favored orientation for the polar moiety of the ligand at the surface of the enzyme. Association free energies range from 2.0 kcal (ATP) to 7.0 kcal (FMN), and the complete quenching of FMN fluorescence on binding suggests that the stacking interaction is quite strong. Calorimetric analysis of FMN binding reveals that both ΔH and ΔS° of association are significant. An unusual feature of the interaction is the strong temperature dependence of ΔCp. Our analysis of the complex demonstrates that ΔGd is not a simple function of the loss of accessible surface for protein and ligand upon binding. On the other hand, loss of accessible surface of the heterocyclic ring system alone correlates directly with ΔGd. We conclude that the association energy derives solely from attractive dispersion forces in which both enthalpic and entropic contributions are significant. Accordingly, change in accessible surface on binding reflects both contributions to the free energy of binding.