An electrostatic model is developed to explain the recently characterized ligand field splittings observed in the core level photoelectron spectra of main group compounds. As for the nuclear electric field gradient splittings observed by Mössbauer and n.q.r. spectroscopy, we show that the electronic splittings also originate from the asymmetric part of the ligand field. Moreover, this ligand field can be divided into the two terms analogous to those used to describe the nuclear electric field gradient splitting: the valence term,
eq
v
, due to the non-uniform population of the valence p, d or f orbitals on the atom M of interest; and the point charge or ligand term,
eq
1,
due to the non-cubic orientation of ligand point charges about
M
. Other ‘cross’ terms which are not present for the nuclear splitting are assumed to be small. We calculate the ligand term,
eq
1
, for the alkali and halide outer p orbitals in the alkali halides, the T1 5d orbitals in TlCl, and the Au 4f orbitals in AuCl-
2
. Wherever experimental results are available, our calculations are in reasonable agreement. The splittings due to
eq
v
for a large number of p, d and f levels are then calculated using a 'pseudo-atomic’ approach with one adjustable parameter — the excess (or deficient) valence orbital population along the
z
-axis, Ap. The two terms are combined to calculate the core level splittings in Me
2
Zn, ZnCl
2
, Me
2
Cd and XeF
2
. Nuclear electric field gradients in these compounds are then calculated from the electronic splittings, and shown to be generally in reasonable agreement with experiment. The importance of open shell Sternheimer shielding—antishielding parameters on both the core electronic splitting and the nuclear splitting is explored and justified.