Agronomists aim to design or co-design agricultural systems that help farmers better meet the challenges of sustainability. In that aim, they need to adapt the way they design to respond to farm diversity, and to deal with knowledge gaps and uncertainty. This study sets out to improve understanding of design reasoning in agronomy. Our assumption is that tracking different reasoning patterns to their roots will shed light on conditions of their development and provide a better understanding of how new agronomic approaches emerge. To do so, we coupled a genealogical approach with a characterization of design regimes. Our work is based on a body of published work by French agronomists from the 18th century to the 21th, selected for their design-oriented nature. Applying an analytical framework built on design theory, we characterized five archetypical design regimes that emerged in connection with developments in the organization of agricultural R&D and shifts in agronomy's epistemological trends. For each one, we described an archetype of agronomists' reasoning, as well as the organizational model and performance logics that conditioned its development. Our results show first that the chains of cognitive operations that make up the agronomists' reasoning are different in each regime, as for example the way agronomists define the design issue and how they explore innovative concepts. Second, our analysis shows that each reasoning pattern generates a specific kind of output which offer farmers different action capabilities: for example, action rules that farmers simply apply, or design support tools to help farmers design techniques themselves. Comparing the design regimes, we identified four factors that influenced agronomists' design explorations: (i) their representations of agriculture, (ii) the scientific tools they use, (iii) their interactions with farmers, and (iv) their links with other scientific disciplines. We finally discuss the contribution of design science to better understanding of the diversity of design reasoning patterns in agronomy and we bring to light some avenues for future research with a view to enriching agronomists' “design toolbox” and supporting co-design in agriculture. We expect this work to offer a useful perspective to agronomists starting a design oriented work to support the evolution of practices on farm.