West Nile virus is a medically significant emerging pathogen for which there is no effective antiviral therapy. The viral protease encoded by NS2B and NS3 is an attractive target for development of an inhibitor and has been the focus of numerous studies. Most have employed recombinant proteases based on an expression strategy we developed which links the essential hydrophilic cofactor domain within NS2B to the NS3 protease domain by a flexible glycine linker. However, autoproteolysis has been a significant problem associated with this construct. The recently resolved crystal structure of the cofactor bound WNV NS3 protease for example, was found to be truncated by 18 residues at its N-terminus. In this study, the autocatalytic cleavage site was identified and removed along with nonessential regions of the glycine linker and cofactor domain. In addition, the optimal size of the NS3 protease was defined. Based on this optimized construct, a recombinant protease incorporating the full length of NS3 was also successfully expressed and purified. Somewhat surprisingly, comparative analysis of the proteolytic activity of this recombinant with that of the protease domain alone revealed little influence of the C-terminal two thirds of NS3 on substrate binding. These modifications have yielded highly stable and constrained recombinant proteases, which are more suitable than existing constructs for both activity and structural studies.