Winter wheat yield is affected not only by extreme climatic events like long-term drought or large-scale flood, but also by less severe unfavourable weather conditions that are liable to increase with future climate change. In this perspective, the goal of this paper is to assess the impacts of agricultural practices on yield robustness to unfavourable weather conditions. Toward that end, we develop a methodological framework based on a statistical typology of wheat cropping systems and a regression model relating wheat yield variability to an index of weather conditions. This framework is applied to a sample of French wheat-growing farms participating in the pesticide-reduction plan known as Ecophyto. The typology includes six classes of wheat cropping systems differentiated on the basis of two types of rotations (simplified versus diversified) and three types of management practices (high-intensive, inflexible management practices; moderately intensive, flexible management practices; and low-intensive, inflexible management practices). Wheat yield in normal or average weather conditions were significantly greater in the high- and moderately intensive cropping systems compared to the extensive cropping systems. By contrast, very few differences were statistically significant for wheat yield robustness to unfavourable weather conditions. This can be explained by the high variability of the robustness parameter derived from the regression model within each cluster. However, heterogeneity of wheat management practices emerges as a potential lever for increasing wheat yield robustness in the face of unfavourable weather conditions.