Improving the measurement of environmental policy intensity would affect not only the selection of variables in environmental policy research but also the research conclusions when evaluating policy effects. Because direct evaluation is lacking, the existing research usually applies data such as pollutant emission data, or the number of policies to construct proxy variables. However, these proxy variables are affected by many assumptions and different selection criteria, and they are inevitably accompanied by endogeneity problems. In this study, China’s environmental policy is comprehensively collected for the first time, and a machine learning algorithm is applied to evaluate the policy intensity. We provide all the policies issued by the Chinese government from 1978 to 2019 and the quantified intensity for each policy. We also distinguish all policies into three types according to their attributes. This dataset can help researchers to further understand China’s environmental policy system. In addition, it provides a valuable dataset for related research on evaluating environmental policy and recommending actions for further improvement.