Significance Peptide binding to MHC receptors is part of a central biological process that enables our immune system to attack diseased cells. We use molecular simulations to illuminate the mechanisms driving stable peptide–MHC binding. Our simulation framework produces an atomistic model of the unbinding dynamics for a given peptide–MHC, which quantifies transitions between the major states of the system (bound, intermediate, and unbound). We applied this framework to study the binding of a SARS-CoV peptide to the HLA-A*24:02 receptor. This work revealed the unexpected importance of peptide’s position 4 in driving the stability of the complex, a finding with broader biomedical implications. Our methods can be applied to other peptide–MHC complexes, requiring only a 3D model as input.