Radioactive C-14 is formed as a by-product of nuclear power generation and from operation of nuclear fuel reprocessing plants like AREVA-NC La Hague (North France), which releases about 15 TBq per year of C-14 into the atmosphere. Since the autumn of 2006, C-14 activity concentrations in samples from the terrestrial environment (air, grass and soil) have been monitored monthly on grassland 2 km downwind of the reprocessing plant. The monitoring data provides an opportunity to validate radioecology models used to assess C-14 transfer to grassland ecosystems. This article compares and discusses the ability of two different models to reproduce the observed temporal variability in grass C-14 activity in the vicinity of AREVA-NC La Hague. These two models are the TOCATTA model which is specifically designed for modelling transfer of C-14 and tritium in the terrestrial environment, and PaSim, a pasture model for simulating grassland carbon and nitrogen cycling. Both TOCATTA and PaSim tend to under-estimate the magnitude of observed peaks in grass C-14 activity, although they reproduce the general trends. PaSim simulates C-14 activities in substrate and structural pools of the plant. We define a mean turn-over time for C-14 within the plant, which is based on both experimental data and the frequency of cuts. An adapted PaSim result is presented using the 15 and 20 day moving average results for the C-14 activity in the substrate pool, which shows a good match to the observations. This model reduces the Root Mean Square Error (RMSE) by nearly 40% in comparison to TOCATTA. (c) 2012 Elsevier Ltd. All rights reserved.