The process of economic development is embodied by rising output per agricultural worker and the exit of labor from agriculture to other sectors, which together result in rising incomes and falling incidence of poverty. This paper addresses agricultural technology's role in slowing or speeding exits of labor from agriculture in Tanzania through its effects on labor productivity. Using a structural multinomial model of occupational choice, I model households' decisions to participation in different activities, i.e. farming, self employment, and working as a wage laborer. Using nationally representative household survey datasets from Tanzania, I jointly estimate households' returns to participating in these activities and their occupational choices conditional on important contextual features, such as agro-climatic potential and market access. Then, I simulate the impacts of interventions, such as improved agricultural technology, infrastructure, and human capital, on labor productivity and labor exits from agriculture. I discuss the implications of the research for policy makers seeking to understand how key policy interventions are likely to influence structural change trajectories in different contexts.