The Chinese government promulgated a series of policies on air cleaning due to the severe PM 2.5 pollution. Thereinto, the Air Pollution Prevention and Control Action Plan (the Action Plan) numbers among the toughest-ever environmental protection policy in China, which significantly restrained haze pollution with a huge cost. In a policy-making for air cleaning, cost assessment is crucial. However, current main methods have a common shortcoming for which they neglect the cost on industrial restructurings. Thus, in this work, we proposed a novel approach to assess the comprehensive cost on the long-term PM 2.5 control in China under the Action Plan scenario. First, PM 2.5 pollution abatements and 10 precursor variations in China’s 31 provinces were quantified via the Cohen’s index. Then, provincial expenditures were jointed with their corresponding Cohen’s indices of precursors and PM 2.5 mitigation, and patterns of the Action Plan were identified through machine learning. It was found that boiler regulation and industrial restructuring were the driving forces for most precursor decreases. Finally, the overall PM 2.5 pollution abatement was calculated by the fixed-effect model and jointed with the cumulative national expenditure to construct a cost curve, which can estimate the cost on both front- and end-of-pipe sections. Thus, the maximum budgets for several air quality targets were evaluated, which was found that China has to pay 10.24 and 51.55 trillion CNY within the PM 2.5 concentrations limits of 25 and 10 μg/m 3 , respectively. Our study provided a new insight of fund gaps between different control targets in China and filled the lack of current cost assessment methodologies.