The hepatitis C virus (HCV) NS3 protease is essential for viral replication. It has been a target of choice for intensive drug discovery research. On the basis of an active pentapeptide inhibitor, 1, we envisioned that macrocyclization from the P2 proline to P3 capping could enhance binding to the backbone Ala156 residue and the S4 pocket. Thus, a number of P2 proline-based macrocyclic alpha-ketoamide inhibitors were prepared and investigated in an HCV NS3 serine protease continuous assay (K(i*)). The biological activity varied substantially depending on factors such as the ring size, number of amino acid residues, number of methyl substituents, type of heteroatom in the linker, P3 residue, and configuration at the proline C-4 center. The pentapeptide inhibitors were very potent, with the C-terminal acids and amides being the most active ones (24, K(i*) = 8 nM). The tetrapeptides and tripeptides were less potent. Sixteen- and seventeen-membered macrocyclic compounds were equally potent, while fifteen-membered analogues were slightly less active. gem-Dimethyl substituents at the linker improved the potency of all inhibitors (the best compound was 45, K(i*) = 6 nM). The combination of tert-leucine at P3 and dimethyl substituents at the linker in compound 47 realized a selectivity of 307 against human neutrophil elastase. Compound 45 had an IC(50) of 130 nM in a cellular replicon assay, while IC(50) for 24 was 400 nM. Several compounds had excellent subcutaneous AUC and bioavailability in rats. Although tripeptide compound 40 was 97% orally bioavailable, larger pentapeptides generally had low oral bioavailability. The X-ray crystal structure of compounds 24 and 45 bound to the protease demonstrated the close interaction of the macrocycle with the Ala156 methyl group and S4 pocket. The strategy of macrocyclization has been proved to be successful in improving potency (>20-fold greater than that of 1) and in structural depeptization.