Although increasing agricultural production is necessary to feed a growing population and meet changing dietary preferences, basing this on expanding area cultivated at the expense of the forest is unsustainable. Expanding agriculture area into forests accounts for 80% of the deforestation globally. Zambia is estimated to lose between 167,000 and 300,000 ha of total forest per annum. Deforestation contributes to climate change, which in turn disproportionately affects smallholder farmers who depend on rainfed agriculture and yet have the least means to adapt to and cope with climate shocks. Climate-smart agriculture (CSA) is considered a necessary condition to increase agricultural productivity and resilience, as well as to adapt to and mitigate climate change. However, the pathways through which CSA can reduce deforestation are neither obvious, nor are they well understood. At conceptual level, the Borlaug hypothesis postulates that increasing agricultural productivity enables intensification, which in turn spares nature. However, increasing agricultural productivity makes agriculture profitable, which in turn might incentivize rather than reduce deforestation—a phenomenon called the Jevons Paradox. Understanding the different conditions and enabling environments for either of the opposing outcomes in different contexts remains an unresolved and important empirical regularity. This paper aims to contribute towards a better understanding of the linkages among CSA, cropland expansion, and deforestation. It unpacks how, why, and where cropland expansion is occurring among smallholder farmers in Zambia.