The spatial-temporal pattern of PM_2.5 in Chinese prefecture-level cities over 1998~2016 has been explored in this study based on Kernel Density Estimation (KDE), Global Spatial Autocorrelation, Local Spatial Autocorrelation, Hotspot Analysis and other methods with the data of PM_2.5 concentration extracted from the 1998 to 2016 global PM_2.5 concentration raster dataset at prefecture-level city. The results showed that first the concentration of PM_2.5 in China has been rising with an average annual increase of 0.55μg/m³; as a whole from 1998 to 2016. Its changing trend could be divided into two stages:rapid growth from 1998 to 2007 and decline-growth-decline fluctuations from 2008 to 2016. Geographically, eastern China and central China experienced similar changing trends, while western China experienced a moderate changing trend, and the changing trend of northeastern China was more dramatic. Second, from 1998 to 2016, the peak of the core density curve gradually shifted to the right, indicating that PM_2.5 pollution in Chinese prefecture-level cities was generally increasing. The shifting was far more substantial in eastern and central China than in western China. Third, PM_2.5 pollution presented a significant positive spatial correlation. The high-value clusters were concentrated in most parts of Shandong, Henan, Hebei, Jiangsu, Anhui, Hunan, Hubei, and eastern Sichuan. The number of high-value cluster cities had increased from 1998 to 2007 with the peak achieved in 2007. The high-value areas westward and southward during this period. After 2007, the number of high-value agglomerated cities had gradually decreased, and its southern boundary gradually moved northward. Low-value city clusters were concentrated in Inner Mongolia, northwestern Heilongjiang, Xinjiang, Tibet, Taiwan, Hainan, Fujian and other regions, and the number of low-value cluster cities showed an increase first and then decreased trend with annual variations.