China’s Environmental Protection Tax, implemented in 2018, permits local governments to adjust water pollutant tax rates, yet existing frameworks largely depend on environmental carrying capacity indicators that are conceptually broad and difficult to quantify. This study advances an innovative approach by integrating water environment self-purification capacity into tax optimization, offering a measurable and ecologically grounded basis for policy design. A dynamic CGE model is developed with an embedded self-purification module to capture the interactions among natural purification processes, pollutant taxation, and regional economic performance. Using the Inner Mongolia River Basin as a case study, simulations for 2021–2030 assess the impacts of alternative tax scenarios. Results reveal three key insights: (1) pollutant tax rates are inversely related to self-purification capacity, providing a quantifiable mechanism for regional tax differentiation; (2) higher tax rates promote emission reduction, though with diminishing marginal effects; and (3) excessive taxation imposes economic costs, affecting industrial output, trade, and consumption. By aligning tax design with ecological capacity, optimized schemes not only improve water quality but also enhance resource balance and support coordinated eco-economic development. The study contributes a novel methodological framework that embeds ecological functions into fiscal policy and provides evidence for adaptive environmental tax reform.