Analysis of available models of equiaxed grain growth in polycrystalline UO 2 nuclear fuel demonstrates the discrepancy between their predictions and experiments under irradiation conditions. As the discrepancy is a systematic overprediction it could be explained if an additional retardation effect exists under irradiation, which is not taken into account by the correlations. As a possible reason of the effect we consider the defect areas arising on the grain faces as a result of interaction with fission tracks. An additional retarding force grows with the irradiation power. Predictions of the model based on the above hypothesis are in qualitative and quantitative agreement with existing experimental data in a wide range of temperatures and intensities of irradiation as well as of fuel burnup.