In computational materials science, many atomistic methods hinge on an interatomic potential to describe material properties. In alloys, besides a proper description of problem-specific properties, a reasonable reproduction of the experimental phase diagram by the potential is essential. In this framework, two complementary methods were developed to fit interatomic potentials to the thermodynamic properties of an alloy. The first method involves the zero Kelvin phase diagram and makes use of the concept of the configuration polyhedron. The second method involves phase boundaries at finite temperature and is based on the cluster variation method. As an example for both techniques, they are applied to the Fe–Cu, Fe–Ni and Cu–Ni systems. The resulting potentials are compared to those found in the literature and are found to reproduce the experimental phase diagram more consistently than the latter.