The present study reports the development of a new procedure for teh theoretical computation of enzyme-substrate specificities. The immediate goal has been to identify experimental data with which computations may be effectively compared, examine the underlying theoretical principles, and demonstrate feasibility. The experimental systems treated are hydrolyses catalyzed by chymotrypsin of Ac-Trp-NH2, of Ac-Phe-NH2, and of the Hein-Niemann "locked" substrate derived from phenylalanine; this may be designated as Lock-HN-OCH3. For Trp and Phe, the L enantiomers are substrates while the D enantiomers are inhibitors, thus indicating differences of 7 kcal/mol or more in delta delta G (D-L). For the "locked" substrate, the D enantiomer is the better substrate and delta delta G (D-L) is -4 to -6 kcal/mol. We have used molecular mechanics to compute steric energies of models for the transition state for these hydrolyses and have been able to reproduce the experimental delta delta G values surprisingly well even with a relatively primitive model. The differences in computed steric energies are not due to any one major term but are rather the consequences of summations of a large number of small terms. The new method shows promise of developing into a useful probe for the quantitative study of biochemical systems.