Fifteen molecules containing the Au(I) species have been calculated by ab initio HF and MP2 methods and by five different density functional approaches. The aurophilic Au(d(10))-Au(d(10)) bonding mechanism has been investigated. Both, one-electron interactions (i.e., electrostatic, polarization, charge transfer, and orbital interference) and two-electron effects (i.e., correlation, dispersion) contribute significantly to the so-called 'secondary' or metallophilic bonds representing the Au-Au interaction. Second, the applicability of density functional approaches to this type of bonding has been tested. It is well-known that present day density functionals are not yet designed to simulate the long-range London dispersion forces between nonoverlapping systems, whereas they approximately reproduce the short range dynamical electron correlations of strongly overlapping chemically bonded nonclegenerate species. It is found here empirically for the investigated groups of gold(I) cluster compounds that simple local density functionals (LDF) of the Slater (or Slater plus Vosko) type yield rather reasonable estimates for the equilibrium distances, and (on the average) also for the aurophilic interaction energies, though with rather large standard deviations. Still LDF are useful for survey investigations of Au cluster compounds. Common gradient corrected DF are not recommended here, nor are the large core pseudopotentials for Au.