Abstract Global optimization approaches are proposed for addressing both the protein folding and peptide docking problems. In the protein folding problem, the ultimate goal involves predicting the native protein conformation. A common approach, based on the thermodynamic hypothesis, assumes that this conformation corresponds to the structure exhibiting the global minimum free energy. However, molecular modeling of these systems results in highly nonconvex energy hypersurfaces. In order to locate the global minimum energy structure on this surface, a powerful global optimization method, αBB, is applied. The approach is shown to be extremely effective in locating global minimum energy structures of solvated oligopeptides. A challenging problem related to protein folding is peptide docking. In addressing the peptide docking problem, the task is not only to predict a macromolecular-ligand structure but to also rank the binding affinities of a set of potential ligands. Many methods have used qualitative descriptions of the macromolecular-ligand complexes in order to avoid the need to perform a global search on the nonconvex energy hypersurface. In this work, a novel decomposition based approach that incorporates quantitative, atomistic-level energy modeling and global optimization is proposed. This approach employs the αBB global optimization method and is applied to the prediction of peptide docking to the MHC HLA-DR1 protein.