Industrial pollution, as one of the major issues closely related to human health, social progress and sustainable development, has drawn a great concern for global vision. This paper uses the methods of coefficient of variation, Theil index, EDSA and SDM model to analyze the spatio-temporal differentiation and the influencing factors of urban industrial pollution (UIP) in China from region-urban agglomeration-urban scales, aiming to provide reference for formulating environmental policies and promoting the development of ecological civilization. The results show that: (1) China's UIP index was declining in fluctuations in 2005-2015, but the overall inequality presented an expanding trend. The industrial pollution index respectively presented the features of "high in the east and low in the west, clustering, and grade-difference" at the regional, urban agglomeration, and urban scales, and the pollution difference of each scale was significant. The gravity center curve of UIP was biased towards the east and showed a southward shift. (2) China's UIP types had transferred from simple shift to complex shift. The pollution pattern of "Eastern>Central>Northeast>West" remained unchanged at regional scale. Urban agglomeration-scale pollution exhibited a pattern of transition from north to south and from inside to outside. Urban-scale pollution presented an evolutionary pattern of shrinking from the center to the periphery and fragmentation to concentration distribution, and it had a certain spatial dependence. (3) There were differences in the influencing factors of industrial pollution at multi-scales. Urbanization rate and industrial structure had significant effects on industrial pollution at the three scales, population density, energy intensity and environmental regulatory intensity, which all had significant impact on industrial pollution at regional and urban scales, the level of industrial development and foreign capital only had significant influence on industrial pollution at regional scale, while the level of science and technology had no significant impact on industrial pollution at the three scales. Finally, we put forward some countermeasures and suggestions to reduce the UIP emissions.

Keywords： multi-scales ; industrial pollution ; spatio-temporal differentiation ; influencing factors