Modelling can be a powerful tool to help us understand and evaluate farmers' practices. It can provide us with references that can then be compared to farmers' practices. We applied this type of modelling approach in order to understand and to evaluate N fertilisation in apple orchards. We used Epistics, an agronomic model that simulates the N and water dynamics in apple orchards and that generates N fertilisation schedules by using a supply demand approach over fixed planning horizons. A schedule is characterised by a total N amount and an application timing. These types of model-generated schedules were evaluated on nitrate leaching and fulfilment of crop N requirements. Schedule generation was performed with different weightings of these two objectives. Farmers' practices were compared to fertilisation schedules generated by Epistics, either with user-defined or with optimised parameters of schedule generation. Model analysis showed that not taking environmental risks into account led to considerable over-fertilisation (e.g., mean of 141 kg N/ha/yr for year 2003 for all plots) with high leaching. With limited consideration of environmental risks, a sharp decrease of the model-generated total N amount to be applied per ha and per year was observed (e.g., mean of 57 kg N/ha/yr for year 2003 for all plots) without restricting fulfilment of N requirements. Farmers' fertilisation practices were similar to those generated by the model that took limited environmental risks into account. They corresponded, on average, to a moderate over-fertilisation, aimed at fulfilling about twice the crop N requirements. For a given total N amount to be applied per ha and per year, the farmers' different fertilisation timing did not influence N leaching or the fulfilment of crop N requirements. Although the variability of total N amounts applied by farmers is partly unexplained, this approach, which combines the analysis and evaluation of farmers' practices, contributed to understandi