As one of the pillar industries, the non-ferrous mining industry has caused severe environmental problems while supporting the development of the national economy. Understanding the spatial pattern and driving force of non-ferrous mining industrial sites (NMISs) is of great significance in promoting the optimization of industrial layout and the overall control of environmental pollution. However, the current research on the evolution of the national distribution pattern of NMISs is still insufficient, which is challenging to meet the actual needs of the existing related industry and environmental protection situation. In this paper, we obtained a high-resolution NMISs dataset, based on multi-source information fusion, including geographic big data, Gaode POI, and special environmental data. We also investigated the spatio-temporal pattern, evolution characteristics, and driving factors of NMISs from the beginning of the 20th century to 2019 based on the spatial regression model and the GIS platform. The results showed that the growth of NMISs in China has generally experienced a stable and slow development trend in the early period (before 1978), gradually reaching a peak after the reform and opening up (1979-2006), and then stabilizing again (2007-). With the continuous enhancement of spatial agglomeration of NMISs, the hotspot areas gradually extended from southeast Hunan and central Yunnan to central and western China, which is rich in resources and energy, presenting an agglomeration pattern of "four cores and multiple sub-cores". The cores include eastern Yunnan, the Hunan-Jiangxi-Guangdong junction area, southern Anhui, and western Henan. The sub-cores included parts of Northwest and Northeast China. Further, the above spatio-temporal evolution characteristics were controlled by the positive promotion of resource endowment, path dependence, and the earlier encouraging industrial policy. However, we observed the emerging negative restraining effect on the recent pattern of NMISs from the tightening of related restrictive industrial and environmental policies introduced intensively after 2010. This paper could provide a reliable scientific basis and information support for optimizing related macro-strategic decision-making and environmental spatial governance by analyzing spatio-temporal patterns and the driving factors of China's NMISs. Furthermore, this study proposed the methodological system for constructing national-scale high-precision industrial site datasets by applying multi-source geographic big data technology. The current paper also provided a new perspective and ideas for the related assessment at a large regional scale.