[en] Arguments are presented, based on the spectroscopic theory of the ionicity, that metal oxides should in most cases be treated as covalent compounds with regard to their crystal radii. It is shown that most 5f fluorite-type oxides should not be stable in the ionic picture. A covalent lattice model for these oxides is proposed and the tetrahedral covalent radius of oxygen is deduced. A discussion of the covalent lattice geometry and of some geometrical aspects of point defects and of dislocations in UO₂ and PuO₂ is given. (author)