Declining soil fertility and increasing rural poverty are major problems facing sub-Saharan agriculture. Bio-economic modeling has been used to analyze the complex interaction between ecological sustainability and rural poverty as well as to explore policy options promoting sustainable development. This paper shows that these models can be further advanced by adopting an agent-based modeling approach. This gives a more realistic representation of diversity in socioeconomic and biophysical terms, allows local interaction between households, and can yield an ex-ante assessment of the distributional consequences of policy intervention. This paper describes the modeling approach and illustrates it with an empirical application to two village communities in the Lake Victoria Crescent of Uganda. It is shown how the model system can be calibrated with and validated against empirical data. The model is used to analyze the potential effect of short-term credit, mineral fertilizer, and improved maize seed on poverty and sustainability. Simulation results suggest substantial reductions in poverty although the incidence of poverty would remain high and these innovations alone would have little effect on the longterm ecological sustainability of the system.