The vapor–liquid coexistence properties of mercury are determined from molecular simulation using empirical intermolecular potentials, ab initio two-body potentials, and an effective multibody intermolecular potential. Comparison with experiment shows that pair-interactions alone are inadequate to account for the vapor–liquid coexistence properties of mercury. It is shown that very good agreement between theory and experiment can be obtained by combining an accurate two-body ab initio potential with the addition of an empirically determined multibody contribution. As a consequence of this multibody contribution, we can reliably predict mercury's phase coexistence properties and the heats of vaporization. The pair distribution function of mercury can also be predicted with reasonable accuracy.