Conservation agriculture is a real alternative in some areas of Madagascar for poverty alleviation and a potential issue for local farmers to overcome climatic and economic risks. Local development projects such as BV-lac in the Lake Alaotra region promote such systems according to local constraints. These days, i.e. 2007-08 cropping season, around 2000 farmers are practising CA cropping system based on : no till permanent organic cover rotation including cover crops. The BV-lac project is currently trying to extend CA cropping system adapted to upland or paddy field. The BV Lac project is trying to enhance extension in quantity and also in quality. One of the major steps to complete this goal is to ensure that technical recommendations matches farm structure and fits smallholder' strategies, constraints and objectives. To tackle such a challenge, we propose an approach including modelling at both cropping and farming systems level. This poster aimed at presenting this approach, the modelling tools and the effective use of such tools with local development partners. Within the modelling tools there are three softwares : the couple of software PRACT/GANESH (K Naudin/SCRID) to select cropping patterns and their integration into farming systems and farming system modelling with the tool OLYMPE (Attonaty, INRA-CIRAD-IAMM). This last one is used in farming systems monitoring network. The PRACT/GANESH tools are still under construct. There are both Decision Support Systems (DSS) PRACT (Prototyping Rotation and Association with Cover crop and no Till) is a tool to generate and to evaluate crop rotations and crop sequences in DMC systems at field level. The second DSS named GANESH (Goals oriented Approach to use No till for a better Economic and environmental sustainability for SmallHolders) is at farm level and uses linear programming to optimize the cropping plan to be proposed to farmers. Olympe is an economic modelling tool on farming simulation to help individual decision-making at farm level an may be used for project to identify technologies adapted to farmers' types. Positive or negative externalities can also be integrated. Farming system modelling associated with a farm typology can therefore be used to help projects as well to test scenarios with various types of technologies in order to assess what is the right technology with the right farmer at the right time. Then, it aims to provide guidelines for agricultural and development policies for institutions or donors.