Crop diversification has a long history in Africa, as a foundation for more resilient and sustainable farming systems. However, success has often been mixed. Variable weather and changing climate requires a focus on supporting farmer capacity to adapt and innovate. Participatory research and simulation modeling are uniquely suited to this goal. Here we present a case study from Northern Malawi where crop modeling in conjunction with participatory approaches were used to evaluate the performance of the promising mixed cropping systems, involving maize and pigeon pea. Using historical rainfall records, simulated yield (Agricultural Production Systems Simulator, APSIM) from maize and pigeon pea-maize intercrop and rotation systems was compared to food requirements for 12 households selected to represent a range of wealth status.Wefound that pigeon pea-maize intercrops were highly likely to produce sufficient calories for smallholder households across variable rainfall patterns, from 73 to 100% of the years simulated, for 10 out of 12 case study households. This stands in contrast to monoculturemaize,where sufficient calories were consistently produced for only half of the case study households. Survey data from this case study documented adoption patterns that reflected strong interest in legume mixtures, and gains in farmer capacity. Farmers shared agronomic information and seeds of pigeon pea and other improved legumes. Overall, we found that farmers were highly motivated to experiment with and adopt legumes that produced food and other valued combinations of traits, whereas green manures met with limited interest. Notably, farmers prioritized species that were reliable at producing food under variable rainfall. Support for farmer-to-farmer learning was critical to the success of the project, and a co-learning approach provided valuable insights to researchers regarding which technologies were more adaptable, and ultimately, adoptable by smallholders living in a highly variable environment.