A cluster expansion is used to predict the fcc groutnd states, i.e., the stable phases at zero Kelvin as a function of composition, for alloy systems. TFile internetallic structures are not assumed, but derived rigorously by minimizing the configurational energy subject to linear constraints. This ground state search includes pair and multiplet interactions which spatially extend to fourth nearest neighbor. A large number of these concentration-independent interactions are computed by the method of direct configurational averaging using a linearizedmuffin- tin orbital Hamiltonian cast into tight binding form (TB-LMTO). The interactions, derived without the use of any adjustable or experimentally obtained parameters, are compared to those calculated via the generalized perturbation method extention of the coherent potential approximation within the context of a KKR Hamiltonian (KKR-CPA-GPM). Agreement with the KKR-CPA-GPM results is quite excellent, as is the comparison of the ground state results with the fcc-based portions of the experimentally-determined phase diagrams under consideration.