In this paper we evaluate the efficiency of extension programmes in the adoption of chemical fertilisers in Ethiopia between 1994 and 2004 using matching techniques in a quasi-experimental setting. Additionally to common factors, measures of culture, proxied by ethnicity and religion, aim to control for potential tensions between extension agents and peasants. We find a considerable impact of extension service on the first fertiliser adoption. Although, there are several exercise of evaluation of extension services policies, this paper manages to disentangle the policy treatment from other confounding effects for two reasons. First we focus on the first adoption decision to isolate the effect of new information provided by the extension service. Secondly, we restrict the sample to villages characterized by late adoption, id est villages where adoption had not occurred before and its appearance can be associated with the policy action with more certitude.