We report a revision of the rigid-ion potential for UO2 introduced by Sindzingre and Gillan (see ibid., vol. 21, p. 4017, 1988). In its original form, the potential is unstable at the reported lattice parameter. We have re-fitted the potential to remove this problem, whilst keeping as close as possible to the original parametrization. The calculated static quantities are little changed, but there is a small increase in the O Frenkel pair formation energy. In dynamics both the superionic transition temperature and the rate of O diffusion in the superionic state are altered, but there is no effect on the calculated thermal conductivity. The conclusions of previous work with the original potential are not affected by the changes we propose.