In our continuing effort to design novel thrombin inhibitors, a series of conformationally constrained amino acids (e.g. alpha-alkyl, N-alkyl cyclic, etc.) were utilized in a systematic structure-activity study of the P3, P2, and P1 positions of tripeptide arginal thrombin inhibitors. Early examples of this effort include: D-MePhe-Pro-Arg-H (15), Boc-D-Phg-Pro-Arg-H (18), D-1-Tiq-Pro-Arg-H (23, D-1-Tiq = D-1,2,3,4-tetrahydroisoquinolin-1-ylcarbonyl), and Boc-D-Phe-Pro-Arg-H (25).10a,20 The current work clarifies the contribution of each residue of the tripeptide arginals toward the potent and selective inhibition of thrombin relative to that of t-PA and plasmin. The alpha-methylarginal modification in the P1 residue resulted in analogs 30 (D-MePhe at P3) and 32 (D-1-Tiq at P3) which had lower potency toward thrombin while exhibiting improved selectivity. Analogs modified at the P2 site were found to be very sensitive to the conformational changes induced by variations in side chain ring size with the flexible pipecolinic acid 31 being 2 orders of magnitude less potent at thrombin inhibition than the conformationally constrained azetidine analog 20. Examination of the P3 binding region indicated that alpha-alkylphenylglycine residues resulted in a tendency to exhibit substantial improvements in selectivity over the nonalkylated residues. Combinations of optimal P3 and P2 changes led to compounds TFA-D-Phg(alpha Et)-Azt-Arg-H (16), TFA-D-Phg(alpha Me)-Azt-Arg-H (17), Ac-D-Phg(alpha Me)-Azt-Arg-H (21), TFA-D-Phg(alpha Me)-Pro-Arg-H (27), 30, and 32, which are clearly more selective for thrombin versus plasmin than the nonconformationally constrained compounds.