Noncovalent interactions are important in many physiological processes of complexation which involve all components of the living cells. Here we report an approach to computationally study the interaction free energies in protein-protein complexes which allows from a single simulation an estimate of the individual contribution of each residue to the binding. We developed this new technique-computational alanine scanning-and applied it to study the interactions of the oncoprotein Mdm2 to the N-terminal stretch of tumor suppressor protein p53. Excellent agreement has been found between the calculated and experimental data. This residue mutation methodology could prove to be a useful general design tool for molecules-nucleotides, peptides, lipids, or any other organic compound-optimized for interactions or stability, since one can qualitatively estimate the free energy consequences of many mutations from a single molecular dynamics trajectory.