Recent advances in the application of kinetic isotope effects to enzyme-catalyzed reactions have provided reliable information for enzymatic transition state structures. A method is presented for quantifying the similarity of substrates and inhibitors with their enzyme-stabilized transition states. On the basis of transition-state stabilization theory for enzymatic reactions, molecules most similar to the transition state structure bind with greatest affinity. Molecular similarity measures are applied to compare substrates, competitive inhibitors, and transition state inhibitors with the transition state structures stabilized by the enzymes AMP deaminase, adenosine deaminase, and AMP nucleosidase. (R)- and (S)-Coformycin 5‘-phosphate are inhibitors for AMP deaminase, with the R-species superior to its enantiomer. Formycin 5‘-phosphate 4-aminopyrazolo[3,4-d]pyrimidine-1-ribonucleotide, and tubercidin 5‘-phosphate inhibit AMP nucleosidase. The transition state for adenosine deaminase is analogous to that for AMP deaminase, allowing analysis of the tight-binding hydrate of purine ribonucleoside and of a weaker inhibitor, 1,6-dihydropurine ribonucleoside. The basis for ranking molecules for similarity to the transition state is the distribution of electrostatic potential at the molecular van der Waals surface. Spatial properties of a molecule are described through the topography of the surface, while the electrostatics capture ionic, hydrogen-bonding, and hydrophobic features. A test molecule is compared with the transition state by orienting the two species so that their van der Waals surfaces are maximally coincident. At this orientation, a single measure sensitive both to the electrostatic potential and its spatial distribution is used to rank the electronic similarity. For AMP deaminase, adenosine deaminase, and AMP nucleosidase, the transition state inhibitors are quantitatively more similar to the transition states than are the substrates. A strong correlation between the binding free energies and the similarity measures is found for most of the transition-state inhibitors in all three enzyme systems. This method is useful in the logical design of transition state inhibitors and may be applied to similarity searches of chemical libraries.