A field water balance and yield model-PILOTE-based on LAI (Leaf Area Index) simulation was evaluated for corn and durum wheat in Montpellier in the south of France. It was used to simulate the soil water content in the active root zone, yield and LAI etc. It was validated against field experimental data under rain fed and irrigated conditions with two main treatments: conventional tillage (CT) and direct seeding on mulch (DSM). The model inputs were soil, crop, irrigation and daily climatic data. Model simulation resulted in good estimations for CT; however some problems were observed in DSM simulations. The model was made up of a soil module with three reservoirs which vary with root increase and plant module. One of the main features of DSM is the retention of crop residues on the soil surface to preserve soil water for crop growth; so a modification which involves reducing soil evaporation in the shallow reservoir of the model was proposed for DSM. This model modification resulted in a closer correspondence between simulated and observed data in DSM. Statistic indices such as prediction efficiency (PE) and RMSE were used to assess model simulation accuracy. The results show that this operative model satisfactorily simulates yield, LAI and soil water content in the active root zone of the crop on a daily basis for two treatments; for example for durum wheat in DSM the model simulation in comparison with observed data for soil water content had PE=0.98 and RSME=9.02 mm. Since the RSME is low and PE is nearby 1, therefore the model satisfactorily simulated soil water content. Furthermore, model satisfactorily simulated other parameters such as LAI, total dry matter and grain yield; so that it can be used for DSM as well as for CT.