Policy makers and practitioners in Afghanistan tend to favor interventions promoting high value agricultural activities, including dairy production. Given the lack of available research and data in the country, the impact of these interventions on traditional subsistence and semi-subsistence crop-livestock households is unclear. A system dynamics model was built to conduct an ex-ante impact assessment of livestock-based interventions in crop-livestock systems in northern Afghanistan. The dynamic simulation model draws on available data, including household level surveys and government price data to predict the outcome of the interventions. The analysis focuses on endogenously generated dynamics and the interaction between biophysical and economic outcomes. Simulation results show that the initial conditions of the households affect the outcome of the interventions. Forage yield improvements and donations of cows may succeed with higher endowment households, whereas interventions that lower transaction costs for milk sales may be more advantageous for households with fewer assets.