AbstractFarmers’ experimentation has long been identified, but until now, its dynamics has hardly been studied at all. The subject is however gaining new interest, as by enabling farmers to learn new cropping practices on their farms, their experimentation could be a driver of transition towards agroecology. The aim of this study is to understand the multiannual experimental itineraries that farmers follow when they try new agroecological practices on their farms. Sixteen cereal and vegetable French farmers were surveyed to establish how they had experimented with agroecological practices over the last decade. They were questioned on their experimentation objectives, the origins of the practice tested, their observations in the field, and the degree of attainment of their objectives. We defined the concept of experimental itinerary as the chronological and logical combination of annual experimental situations. Similarities and differences between the experimental itineraries of the 16 farmers were sought. The results are threefold. First, we showed how a farmer progressively builds an experimental itinerary over years after the introduction of a brand new practice. Second, by analyzing the 33 experimental itineraries identified in the sample, we demonstrated that some types of experimental situation occur more frequently at the beginning, some in the middle and some at the end of the experimental itinerary. Third, we identified four patterns of experimental itinerary (low investment, linear, tree, and grove patterns) that correspond to different ways the farmers organize the annual experimental situations over the years. The novelty of the results lies in the in-depth understanding of the inter-relations between successive experimental situations that has never been described before. We finally discuss the results with an operational focus, that is, how the representation of experimental itineraries could support farmers in their experiments and enable them to learn more efficiently.