Accurate estimation of water demand at the irrigation scheme scale is a key requirement for water management, which is made difficult by the large diversity of crops and production systems. The main objective of this study was to estimate irrigation water demand at the farm scale, taking different types of knowledge into account: (i) database study, where water demand is supposed to be the supply (L0); (ii) actual cropping pattern (L1),; (iii) actual irrigation techniques and cropping pattern (L2); and (iv) actual irrigation practices, actual irrigation techniques, and cropping pattern (L3). Farm typology makes easy this estimation as it takes into account various farm characteristics. Firstly, a farm typology was established based on 115 farms surveyed in the irrigation scheme of Borj Toumi Toungar (Tunisia), selected for the study. Secondly, climatic crop water irrigation requirements were estimated using the agro-meteorological water balance model Pilote. Typology results were used to estimate water demand at the scale of family farms taking into account different levels of knowledge Six classes of farming areas were identified based on irrigation subsystems, cropping pattern (fruit tree area, cropping vegetable area and field crop area), intensification crop level and cropped surface. Results showed that supply could not entirely cover the climatic water demand. However, it could nearly meet actual water demand at the farm scale. According to the water demand estimation at different levels, results showed that there were few changes in water demand at farm scale when moving from L1 to L2. At the opposite, actual water demand (L3) represented about half the climatic water demand (L2). Calculations based on farm classes highlighted the importance of actual farm practices. Within the same class, the difference between water demand estimation at different levels of knowledge was remarkable. Network rehabilitation is thus necessary to enable farmers to increase water delivered to crops and limit water stress; the collective network was conceived to deliver a flow of 0.34 l/s/ha in this sector. Given changes in farming and varietal choices, this flow proved insufficient to face the climatic water demand.