Based on work reported by several authors, one of us (RP) has introduc
ed a Mulliken-like population analysis of pair densities. This scheme,
introduced first at the semi-empirical level, was subsequently genera
lised both for SCF and for post-SCF ab initio methods. A potential pro
blem is the basis set dependence that can be expected for all kinds of
Mulliken-like approaches. The main purpose of the present study is to
quantify this basis set dependence for a range of simple molecules, u
sing basis sets ranging from STO-3G to TZVP quality. It is shown, exce
pt for low quality basis sets, that the >>effective<< pair populations
derived from ab initio SCF calculations are relatively insensitive to
the choice of basis set.