This paper, seeks to empirically establish the complementarity or otherwise between the decision to adopt improved soybean variety and the decision to participate in extension service training. The paper departs from the traditional binary dependent regression model (probit and logit) with extension treated as one of several covariates and instead model the binary outcomes of the decisions to adopt improved varieties, and access to extension services simultaneously using bivariate probit model. Data for the study is from 1432 farmers across the three regions of northern Ghana to jointly model the determinants of access to agricultural extension services and adoption of improved soybean variety (Jenguma). We found a positive correlation between the decisions to adopt improved soybean variety (Jenguma) and access to agricultural extension services. The implication is that, the decision to adopt the variety is interrelated with access to agricultural extension services. Hence, access to agricultural extension services is complementary to the decision to adopt new soybean varieties. Findings also indicate that, the decision to adopt improved soybean varieties is influenced by younger, less educated farmers with large farms and ownership of parcels. Given that agricultural extension service is a public good, we recommend that the government of Ghana allocates more human, financial, and technical resources to the extension services to enhance the delivery of extension services to farmers so as to improve the adoption of productivity enhancing technologies