In order to solve the problem of cost planning regarding environmental governance, the current solution is mainly based on the time series-based cost forecast methods, and neglects to consider the efficiency of environmental governance. Thus, a method of adaptive network-based fuzzy inference system(ANFIS) with efficiency preference is proposed for cost planning of environmental governance, whose basic idea is to consider the input-output relationship and efficiency in environmental governance by using the data envelopment analysis(DEA) undesired output model and ANFIS, respectively. In the case analysis, according to the environmental governance data of various provinces in China from 2003 to 2015, the performance analysis and comparison are carried out in terms of accuracy and effectiveness to compare with the conventional methods. The results show that the proposed method is effective and has better accuracy than those methods.