Polarized neutron and X-ray diffraction experimental results on simple Cr(III), Ni(II) and Co(ll) complexes are compared with theoretical calculations. A simple ionic (crystal-field) model is a useful first approximation for the description of the spin and charge densities. For the more covalently bound cases the effects of electron-electron correlation in the metal-ligand bonding are as important as the spin and charge transferred via simple covalence. Thus simple M.O. models do not offer much improvement, nor doab initiocalculations which do not include configurational interaction (viz. restricted Hartree-Fock). Unconstrained calculations which partly incorporate electron-electron correlation are in qualitative, but not quantitative, agreement with the experiments. The polarized neutron diffraction experiments provide such sensitive tests of theory because they spatially separate the small effects in the bonding region from the metal atom 3d region, whereas these are inextricably mixed in the energetic properties measured by spectroscopic techniques.