This article examines the determinants and impact of conservation agriculture (CA) technology adoption on farm household welfare in Zambia. To account for selection bias from both observable and unobservable factors, an endogenous switching regression model is employed to estimate the impact of the technology on continuous outcomes like farm output, throughput accounting ratio (TAR), poverty gap, and severity of poverty. A recursive bivariate probit model is however used for the estimation of impact of adoption on a binary outcome like poverty headcount. The empirical findings demonstrate that the adoption of CA technology increases maize output, and farm TAR and reduces household poverty. Moreover, the results reveal that farmers’ years of schooling, social networks, access to credit, extension services, and machinery as well as soil quality positively influence adoption of CA technology.