Farmers in developing countries can dramatically improve their productivity by adopting new plant varieties. Yet, informational barriers often mean adoption rates remain low. In this study, we focus on how learning from others represents one means of removing informational barriers. We capture the effect of social learning through an explicitly spatial econometric model, applied to farm-level maize adoption rates in Mozambique. We find that social learning is significant, and explains the apparent clustering of adoption among farmers. Agencies interested in promoting variety adoption, therefore, would be well-served to leverage the strength of existing information networks, rather than imposing solutions that work against inter-farmer information flow.