Corporate Social Responsibility (CSR) is closely bound up with the supply of quality tourism products, the quality of tourism experiences, and the sustainable development of tourism. The various crisis risks and pressures that CSR failure causes are transmitted through the supply chain. However, tourism CSR research lacks literature on thinking about the decision-making game process between enterprises, from the perspective of the tourism supply chain system. The lack of systematic optimization decision theory leads to the backward practice of tourism CSR. Breaking the boundary of thinking about CSR from a purely static enterprise perspective to place CSR within the framework of a tourism supply chain dynamic operation establishes a new and robust theoretical analysis path for tourism CSR research that promotes matching tourism CSR research and practice. Utilizing different power structures, this paper studies four models of tourism supply chain game decision-making: the Stackelberg game of scenic-spot leadership, the Stackelberg game of travel agency leadership, the Nash equilibrium game of scenic spot and travel agency, and the centralized decision-making model of complete cooperation in a two-level tourism supply chain, comprising a travel agency and a scenic spot considering CSR. Further, through the equilibrium decision comparative static analysis and the example simulation analysis, this paper found that tourism products’ wholesale price, sale price, and demand, the CSR commitment level of the scenic spot, the scenic-spot profit in a Nash equilibrium game, and the profits of the travel agency and the supply chain will increase. With increased CSR effect, the profit of the scenic spot under the Nash equilibrium game presents a dynamic change from increase to decrease as the CSR effect. The CSR commitment level of the scenic spot is highest under the centralized decision-making model, and minimal in the scenic-spot leadership model. When the CSR effect factor exceeds a critical value, the travel agency will respond to scenic spot’s fairness concerns. Hence, the wholesale price of products under the leadership of the travel agency will be higher than under the leadership of the scenic spot. Centralized decision-making is better than decentralized decision-making, and under decentralized decision-making, the optimal profit of the tourism supply chain under scenic-spot leadership is the lowest. That is, the optimal profit of the tourism supply chain under travel-agency leadership and the balance of power on both sides is better than under scenic spots. Therefore, when forming a centralized decision with full cooperation is impossible, having the travel agency in a leading position is most beneficial. Strong CSR effect is also most beneficial for the tourism supply chain system. However, from the perspective of its own profits, the scenic spot’s leading position is not necessarily the most advantageous, and the size of the CSR effect should determine that. Through modeling optimization research, tourism CSR and channel power are both introduced into the decision-making model of the tourism supply chain for the first time. This develops the research perspective and method of tourism CSR and, at the same time, provides some decision support for CSR practice on the tourism supply side, under different power structures.