Evidence is presented that there is a clear covalent component in the bonding of Au+ to Kr and Au+ to Xe, with some evidence that there may be such bonding between Au+ and Ar; for Au+ and Ne, there is no such evidence, and the bonding seems to be entirely physical. A model potential analysis shows that when all attractive inductive and dispersive terms out to R-8 are properly included in the Au+-Ne case, with an Ae(-bR) Born-Mayer repulsive term, essentially all the bonding in Au+-Ne can be rationalized by physical attraction alone. This is consistent with a natural bond order (NBO) analysis of the Au+-Ne ab initio wavefunctions, which shows the charge on Au+ to be very close to 1.0. In contrast, similar model potential and NBO analyses show quite clearly that physical interactions alone cannot account for the large bond energy values for the Au+-Kr and Au+-Xe complexes and are consistent with covalent contributions to the Au+-Kr and Au+-Xe interactions. Au+-Ar is seen to lie on the borderline between these two limits. In performing the model potential analyses, high-level ab initio calculations are employed [CCSD(T) energies, extrapolated to the complete basis set limit], to obtain reliable values of R-e, D-e and omega(e) as input. A comparison of the gold-Xe bond distances in several solid-state Au(I, II and III) oxidation-state complex ions, containing "ligand" Xe atoms, prepared by Seppelt and co-workers, with that of the "free" Au+-Xe gas-phase ion is made, and a discussion of the trends is presented.