HCV drug discovery efforts have largely focused on genotype 1 virus due to its prevalence and relatively poor response to current therapy. However, patients infected with genotype 2 and 3 viruses account for a significant number of cases and would also benefit from new therapies. In vitro studies using two chemically distinct protease inhibitors with clinical potential showed that one, VX-950, was equally active on proteases from all three genotypes, whereas the other, BILN 2061, was significantly less active on genotype 2 and 3 proteases. Naturally occurring variation near the inhibitor binding site was identified based on sequence alignment of the protease region from genotype 1-3 sequences. Substitution of amino acids in genotype 1 based on genotype 2 and 3 has revealed residues which impact binding of BILN 2061. Substitution of residues 78-80, together with 122 and 132, accounted for most of the reduced sensitivity of genotype 2. The most critical position affecting inhibitor binding to genotype 3 protease was 168. Substitution of residues at positions 168, 123, and 132 fully accounted for the reduced sensitivity of genotype 3. Comparative studies of BILN 2061 and a closely related nonmacrocycle inhibitor suggested that the rigidity of BILN 2061, while conferring greater potency against genotype 1, rendered it more sensitive to variations near the binding site. Free energy perturbation analysis confirmed the experimental observations. The identification of naturally occurring variations which can affect inhibitor binding is an important step in the design of broad-spectrum, second generation protease inhibitors.