Sub-Saharan African economies depend heavily on agriculture. Seed from improved varieties and other inputs are imperative to the transformation of the agricultural sector from subsistence farming to small-scale commercial agriculture and thus to increasing food security on the continent. Farmers make the decision to adopt seed from improved varieties based on a number of seed attributes. These range from tangible attributes such as input costs and yield to intangible attributes such as trust in seed from improved varieties. In the course of adoption decisions, social dynamics involving trust can over-ride objective evaluations of tangible attributes. This makes it difficult to design sustainable adoption policies in an intuitive way. For this purpose, we develop a system dynamics model and combine it with conjoint analysis. Conjoint analysis allows us to elicit smallholder farmers’ choice preferences in detail and to add precision to the structure of the model. The simulation framework helps to improve our understanding concerning the dynamic implications of accumulation processes relating to trust and skill. We test this approach with empirical data for maize in Malawi. Model simulations demonstrate that effective adoption stimulation policies should focus on measures that build trust in improved maize varieties instead of increasing their potential yield even further and, in this way, contribute to food security. Copyright © 2012 John Wiley & Sons, Ltd.