<p>A novel bilevel programming model (FCSM) is proposed for China’s cross-border forestry carbon sink (FCS) planning considering regional collaboration strategy, where the central government acts as the leader who pursues the minimization of its development cost by setting carbon sink forests (CSF) subsidy, while the local governments are the followers who weighing the options of expanding their own CSF or engaging in carbon trading with other local governments to minimize their FCS development costs. Considering the inherent complexity, the established bilevel FCSM model is first transformed into a single-level nonlinear program by Karush-Kuhn-Tucker conditions, and then solved by an improved particle swarm optimization algorithm. The case study in Beijing-Tianjin-Hebei region implies that the FCSM model achieves its Stackelberg equilibrium. The area of CSF in Beijing and Tianjin need to be reduced by 43.40% and 24.08% respectively, while Hebei should increase by 13.45%. Notably, the costs incurred by central government in China decrease by 19.84%.</p>