As elsewhere in the world, pesticide reduction is not achieved in France (Hossard et al., 2017) despite an ambitious public policy aiming to foster practice change on commercial farms (the Ecophyto plan). Most of the studies analyzing obstacles to pesticide reduction at the farm level compile generic lists of factors hindering change without showing how these factors interact with one another, or how they relate to farmers' practices and their agricultural context (e.g. Bjørnåvold et al., 2022; Darnhofer et al., 2005). Indeed, these studies mainly emphasize generic constraints linked to the characteristics of alternatives (e.g., their limited availability, low effectiveness, or high labor requirements) or to farmers' motivations and attitudes towards agroecological transitions, but they rarely investigate how farmers’ practices choice, and the agronomic functioning of their cropping systems may impede pesticide reduction. The aim of our study was to explore farmer’s practices and their rationale, in order to decipher the farmer’s logic of pesticide use and unuse. MethodologyWithin the DEPHY-Farm network (a national farmer network enhanced to develop alternative practices to pesticide use), we selected farmers who showed contrasted trajectories of their Treatment Frequency Index (TFI) across years since their entry in the network (increase, reduction or stability at a high level, Table 1). We conducted interviews with 25 farmers in four different regions in France, in order to understand farmer’s practices and the rationale behind them. For each farmer, we produced a narrative using the action logic framework (Quinio et al., 2022; Salembier, 2019) to organize the systemic links established by the farmer during the interview between his/her practices, the interpretation he makes about agronomic processes at play, the information he uses to manage his crops, his satisfaction criteria, and contextual elements. By comparing the logics of farmers with different TFI trajectories, and drawing on agronomic literature, we identified, through the action logics, key systemic mechanisms explaining the high dependency on pesticides, thus depicting a systemic analysis of the obstacles to pesticide reduction on farms.We found that obstacles mentioned by the interviewed farmers varied from one farm to another, and an obstacle for one farm may not be an obstacle for another farm. Most often, obstacles are not related to a particular practice or pest, but are closely interlinked with the way farmers design their cropping systems. For instance, one farmer who prioritizes yield as a satisfaction criterion chooses a late-maturing corn variety and opts for early sowing to extend the crop cycle and therefore maximize yield. He systematically applies an insecticide while sowing the crop, as he reports high wireworm pressure in his plots, which could lead to significant yield losses (obstacle to pesticide reduction). Corn frequently returns in the crop rotation on irrigated plots, as it is considered the most profitable crop by the farmer, who has invested in a corn drying unit that needs to be cost-effective. As a result, he cultivates a large area of corn every year, which make it more difficult to shift sowing dates. By contrast, his neighbor, who had a similar rationale in the past, decided to reduce his corn acreage and to introduce other crops in order to break the cycle of weeds and pests while also spreading the economic risk associated with a single crop. He now selects an early-maturing variety that can be sown later, allowing the crop to establish under conditions where corn is more vigorous and quickly becomes less vulnerable to wireworm attacks. As a result, he applies no insecticides on his corn crops and is satisfied with this cropping system, as it helps reduce input costs (irrigation, pesticides, drying costs). However, he mentions that this logic is more suitable on relatively small farms like his, as a delay in sowing dates is easier for a small maize acreage. In contrast, farmers with large maize areas often need to start sowing as soon as possible, as the operation is spread over a longer period. This example show that pest pressure and pesticide use depend on farmer’s combination of practice (here crop rotation, crop variety and sowing dates), driven by specific satisfaction criteria and contextual elements (here farm size and specialized equipment).Using a cross-analysis of farmer’s action logics, we identified several systemic mechanisms that hinder pesticide reduction at the farm level in arable systems, among which “system simplification to reduce workload” and “cash crops oriented diversification”. We identify two key methodological contribution of this work. First, we used an in-depth analysis of the action logics of farmers who have reduced their TFI to put into perspective the logics of those who have not. In doing so, rather than identifying a list of generic obstacles, we unraveled strongly interlinked mechanisms at the farm level, which alert on certain trends in the agricultural sector (among which the increase of farm size and over-equipment of farms with cutting-edge equipment not well suited to system redesign). Moreover, the action logic framework so far has been used to highlight innovative agroecological practices, through innovation tracking for instance (e.g. Verret et al., 2020). In this study, we used the same concept to highlight systemic barriers to pesticide reduction at the farm level. A possible extension of this work would be to assess whether this systemic analysis could help farmers, agricultural advisors, researchers or other stakeholders to overcome the identified obstacles and design new strategies to support pesticide cut in arable systems.