A new computational approach is proposed to probe the importance of residue side chains for the stability of a protein fold. Computational mutations to estimate protein stability (CMEPS) is based on the notion that the binding free energy corresponding to the complexation of a given side chain, considered as a "pseudo-ligand" of the wild type protein, reflects the importance of this side chain to the thermodynamic stability of the protein. The contribution of a particular side chain to the folding energy is estimated according to the molecular mechanics-generalized born surface area MM-GBSA approach, using a single molecular dynamics simulation trajectory of the wild type protein. CMEPS is a first principles method which does not contain any adjustable parameter that could be fitted to experimental data. The approach is first validated for Barnase and the B1 domain of protein L, for which a correlation coefficient R = 0.73, between experimental and CMEPS calculated DeltaDeltaG values, is found and then applied to the insulin monomer. In the present application, CMEPS replaces each amino acid by an alanine residue. Therefore, most mutations lead to cavities in the protein. From this the change in stability can be correlated with increased cavity volume. For insulin, this correlation is very similar compared with data previously analyzed for T4 lysozyme from an experiment for buried apolar side chains. There, the increased cavity volume has been related to the hydrophobic effect. However, since CMEPS uses the energetics in terms of electrostatic and van der Waals interactions (and not the hydrophobic effect which is difficult to relate to physical interactions), it is possible to study the effect of mutations of polar and solvent accessible side chains. According to CMEPS, residues Leu A16, Tyr A19, Leu B11, Leu B15, and Arg B22 are most important for the stability of the monomeric insulin fold. This is in agreement with experimental data. As a consequence, mutation of these residues may lead to misfolded and inactive insulin analogues.