No-tillage techniques are rapidly developing in France (accounting for 21% of cultivated land in 2001, and 34% in 2006), mostly for economic reasons. These techniques profoundly change the agroecosystem, having both positive environmental effects (increasing biodiversity, combating erosion, increasing carbon reserves etc.) and possible negative effects due to the difficulty of managing such systems (including, in particular, a greater use of herbicides). They also lead to “sea change” innovations, so-called because they call into question accepted knowledge (such as the introduction of cover crops to fulfil agroecological functions). The changes in practices underlying the development of conservation agriculture are based on knowledge (concerning biological regulation, mechanisation) not widely available in the institutes involved in research and development. The implementation of such practices requires several years of learning, close accompaniment and the creation of local references. These changes, which were initially poorly supported by institutional bodies, have led to the emergence of groups of farmers, often organised around company or cooperative representatives, who share their experiences and learn together (Goulet, 2008). These “practices based communities” constitute innovation laboratories trying out new means of generating knowledge and bringing together farmers, local agricultural advisors and, in some cases, researchers. The aim of this communication is to examine the link between the means of producing knowledge and the help provided to farmers within a community bringing together farmers practising conservation agriculture and actors in the domain of development. This should generate insight into the tension between production, the testing of knowledge in these innovative, localised experiments and the need for the knowledge obtained to be generic, translated as soon as it is exported to another site and other people in the framework of sustainable development Our work is original in that it analyses the conversion of communal knowledge into generic knowledge, on the one hand, and the conversion of a participatory process for the assistance of a small group of farmers into a larger process for spreading alternative techniques favouring sustainable development, on the other. The key question is, therefore, how can knowledge generated locally and collectively within a group be transmitted to, and used elsewhere by others? Our work is based on an intervention study of a club established within a cooperative in 2004, with the aim of aiding 50 farmers using no-tillage and conservation agriculture techniques. Since 2008, the cooperative has been following a business plan defined in terms of sustainable development based on the notion of “ecological intensification.” Conservation agriculture is a textbook example of the approach followed. The activities of the club are therefore at the forefront of this work, providing a showcase and a laboratory for the concrete implementation of sustainable development. The key issue for the cooperative is to extend what has been achieved with a small group of farmers and advisers to all 2000 members of the cooperative. To meet this challenge, the leaders of the cooperative have sought support from the research community. This led us to initiate “coaching” activities in 2008, to assist the cooperative with valorisation and the organisation of systems for producing knowledge. It also led us to associate the representatives of the cooperative with our PEPITES research project, which is funded by the ANR. Our approach is interdisciplinary, combining agronomy and sociology. The material used in this work was of two origins: Interactions between advisors, researchers and the leaders of the cooperative during three days of coaching and as summarised in reports. A survey based on comprehensive interviews (Kaufmann, 1996) with half the members of the club. The results obtained were as follows: Community based knowledge was generated in several ways: 1) through exchanges between farmers, with or without advisors; 2) through experimentation, ranging from simple trials without controls to more formal scientific experiments 3) through the combination of different types of knowledge, including knowledge obtained from elsewhere./The knowledge produced by the club was of a pragmatic nature, resulting from action and testing on farms. This knowledge was therefore “localised”, in the sense that the validation of this knowledge was local. Within the club, we identified two different types of farmers on the basis of their approach to the localised nature of this knowledge. The first group of farmers were able to extrapolate from a local situation (soil conditions very different from their own, for example), to generate knowledge useful in their own conditions. The second group of farmers went no further than “that doesn’t apply to me.”. The farmers of the first group were able to translate local knowledge into reasoning based on tha tknowledge, rendering it more generic. The expectations of the two groups concerning accompaniment and monitoring were not the same. The validation and valorisation of locally produced knowledge (on a farm, within the club) further afield requires a formalisation of the knowledge concerned (often in written form) or its incorporation into a technical object. We tested two methods: 1. Stabilising local knowledge and combining it with knowledge from elsewhere, in a collective production experiment. This led to the production of a guide on cover crops which was distributed to all the members of the cooperative. 2. Helping the advisors to move from formalised facts (localised at plot level), to formalised reasoning, particularly during “field” meetings and to write reports on what they have done. The reasoning may be generic, even if it leads to different decisions being taken in different local contexts.