This work is aimed at clarifying the changes on optical spectra of Cr(3+) impurities due to either a host lattice variation or a hydrostatic pressure, which can hardly be understood by means of the usual Tanabe-Sugano (TS) approach assuming that the Racah parameter, B, grows when covalency decreases. For achieving this goal, the optical properties of Cr(3+)-doped LiBaF(3) and KMgF(3) model systems have been explored by means of high level ab initio calculations on CrF(6)(3-) units subject to the electric field, E(R)(r), created by the rest of the lattice ions. These calculations, which reproduce available experimental data, indicate that the energy, E((2)E), of the (2)E(t(2g)(3)) → (4)A(2)(t(2g)(3)) emission transition is nearly independent of the host lattice. By contrast, the energy difference corresponding to (4)A(2)(t(2g)(3)) → (4)T(1)(t(2g)(2)e(g)(1)) and (4)A(2)(t(2g)(3)) → (4)T(2)(t(2g)(2)e(g)(1)) excitations, Δ((4)T(1); (4)T(2)), is shown to increase on passing from the normal to the inverted perovskite host lattice despite the increase in covalency, a fact which cannot be accounted for through the usual TS model. Similarly, when the Cr(3+)-F(-) distance, R, is reduced both Δ((4)T(1); (4)T(2)) and the covalency are found to increase. By analyzing the limitations of the usual model, we found surprising results that are shown to arise from the deformation of both 3d(Cr) and ligand orbitals in the antibonding e(g) orbital, which has a σ character and is more extended than the π t(2g) orbital. By contrast, because of the higher stiffness of the t(2g) orbital, the dependence of E((2)E) with R basically follows the corresponding variation of covalency in that level. Bearing in mind the similarities of the optical properties displayed by Cr(3+) impurities in oxides and fluorides, the present results can be useful for understanding experimental data on Cr(3+)-based gemstones where the local symmetry is lower than cubic.