In response to the diversity of planting structures in tropical regions and under the constraints of total water use control and the construction of agricultural “two zones”, a multi-objective optimization model was developed based on the water footprint theory to maximize both economic and ecological benefits. This model focused on the planting structures of four typical crops (rice, corn, fruit, and sugarcane) in three counties (or cities) in Xishuangbanna Prefecture, Jinghong City, Menghai County, and Mengla County, in the tropical region of Yunnan Province. The NSGA-Ⅱ genetic algorithm was used to solve the model, resulting in two sets of optimized schemes: one that maximizes the net economic benefit per unit of water and the other that minimizes the grey water footprint for optimal ecological benefit. Using the coupled entropy weight method and TOPSIS evaluation method, optimal agricultural planting structures were determined for the three counties (or cities). The optimization results show that reducing the corn planting area and increasing the fruit planting area by 18.8% to 57.4% in Xishuangbanna Prefecture can significantly reduce the grey water footprint and improve the net economic benefit per unit of water. In Menghai and Mengla counties, increasing the rice planting area by 13.6% to 67.7% did not reduce the net economic benefit per unit of water. Sugarcane can have its planting area appropriately expanded in Jinghong City and Mengla County, based on improved agricultural production conditions. After optimizing the planting structure, the net economic benefit per unit of water can increase by 13.0% to 62.0%, and the grey water footprint can decrease by 1.9% to 28.0%.