Abstract: Atmospheric capacity theory has been one of the important fundamental theory for air pollution control policy and management in China for a long time, but it is ambiguous in terms of concept definition and theoretical model and is not able to establish commonly recognized mathematical models and algorithms. Based on the review of the development and application of atmospheric capacity theory, this paper discussed the concept definition, connotation and the difficulties and limitations of the theoretical model and algorithms. The difference between the absolute, generalized and narrowly defined concepts of atmospheric capacity is proposed. The main theoretical difficulties of atmospheric capacity planning are identified, i.e., the openness of the atmospheric planning space and the resulting spatial border definition difficulty and the absence of constraints outside the planning domain. Except for a quasi-confined space such as nocturnal boundary layer or valley complex terrain where generalized atmospheric capacity can be defined for a primary air pollutant, it is unable to obtain a definite optimal capacity solution for most of other cases. Due to the non-one-to-one correspondence and non-linear relationship between secondary air pollutants and their precursors, the traditional capacity theory for single air pollutant specie and single objective linear optimization methods are no longer applicable for ozone and secondary fine particulate air pollutants problems. The atmospheric capacity theory is not a valid and universal theory, especially for regional problems, because air quality problems may no longer be capacity-exceeding problems. For the regional combined air pollution problem, it is recommended to adopt a scenario analysis or impact assessment planning approach using air quality models at the national or regional scale based on the status quo and future emission scenario as well as economic and technical feasibility to obtain the optimized regional emission cap. In the long run, a multi-species and multi-objective planning approach that takes into account technical, economic and social variables could be considered.