Recently, Schweizer, Honnell, and Curro reported a self-consistent formulation of the polymer reference interaction site model (PRISM) theory for polymer melts. The purpose of this paper is to compare the predictions of this theory to Monte Carlo simulations of hard chains thereby allowing an independent test of the various approximations in the PRISM theory. We find that the self-consistent PRISM theory with the Percus–Yevick (PY) closure is very accurate for both the intramolecular and intermolecular correlations in hard 20-mers for a wide range of densities. The accuracy of the theory for longer chains is somewhat diminished. We also compare the PRISM theory predictions using the PY, hypernetted chain (HNC), and Martynov–Sarkisov (MS) atomiclike closures. All the closures give equally good predictions at high densities, but the HNC and MS closures exhibit unphysical features, and ultimately fail to converge, at lower densities and/or large chain lengths.