The adoption of innovations to improve yields and reduce poverty has become an important issue for the agricultural sector. However, in spite of all efforts implemented, the adoption of these technologies remains below the expected levels. This article aims to analyze the effects of communication networks on the adoption and diffusion of improved maize varieties promoted in the North of Benin in order to increase its productivity. Data were collected among maize producers identified with the snowball method in Nikki town. Logistic regression was carried out to analyze the influence of socio-relational, socio-economic and demographic factors on the adoption and diffusion of improved maize varieties. Software R version 3.4 was used for statistical data analysis. The Akaike Information Criterion (AIC) was used to choose the best model for predicting adoption and diffusion. Results show that rich men with a high degree of authority and power tend to have a high eigenvector centrality while men with a high degree of centrality and betweenness centrality tend to have a high level of education, a high area and a high degree of authority and power. In addition, a combination of socio-relational socio-economic and demographic variables predict better the probability of adoption (AIC=52.929) while socio-relational variables predict better the probability of diffusion (AIC=15.819). The study suggests that the diffusion of improved maize varieties at a large scale must involve persons who are rich, with a large agricultural land a good education level, and powers in their locality.