Strict enforcement of forest protection and massive afforestation campaigns have contributed to a significant increase in China’s forest cover during the last 20 years. At the same time, demographic changes in rural areas due to changes in reproduction patterns and the emigration of younger population segments have affected land-use strategies. We identified proximate causes and underlying drivers that influence the decisions of farm households to plant trees on former cropland with Bayesian networks (BNs). BNs allow the incorporation of causal relationships in data analysis and can combine qualitative stakeholder knowledge with quantitative data. We defined the structure of the network with expert knowledge and in-depth discussions with land users. The network was calibrated and validated with data from a survey of 509 rural households in two upland areas of Yunnan Province in Southwest China. The results substantiate the influence of land endowments, labor availability and forest policies for switching from cropland to tree planting. State forest policies have constituted the main underlying driver to the forest transition in the past, but private afforestation activities increasingly dominate the expansion of tree cover. Farmers plant trees on private incentives mainly to cash in on the improved economic opportunities provided by tree crops, but tree planting also constitutes an important strategy to adjust to growing labor scarcities.