Using binding models which were based on the X-ray crystal structure of an amino acid-based active site-spanning inhibitor complexed with cathepsin K, Cbz-leucine mimics have been developed, leading ultimately to the design of a potent cathepsin K inhibitor free of amino acid components. These mimics, which consist of alpha-substituted biphenylacetyl groups in place of Cbz-leucine moieties, effectively mimic all aspects of the Cbz-leucine moieties which are important for inhibitor binding. The predicted directions of binding for the inhibitors were confirmed by mass spectral analysis of their complexes with cathepsin K, which gave results consistent with acylation of the enzyme and loss of the acylhydrazine portion of the inhibitor which binds on the S' side of the active site. The binding models were found to be very predictive of relative inhibitor potency as well as direction of inhibitor binding. These results strengthen the validity of a strategy involving iterative cycles of structure-based design and inhibitor synthesis and evaluation for the discovery of non-peptide inhibitors.