Based on the ERA5 reanalysis data, PM_2.5 monitoring data and meteorological observation data of Yibin from 2015 to 2023, with the principal component analysis in T-mode (PCT) objective classification method, the 850 hPa geopotential height fields during the months with severe PM_2.5 pollution (January, February, November, and December) were categorized into six weather types. The PM_2.5 pollution conditions, meteorological characteristics, and potential source regions under different weather types were analyzed. The results show that: (1) The eastern high-pressure type (Type 3) and low-pressure bottom type (Type 4) were prone to PM_2.5 pollution, while the back of a low vortex type (Type 5) and saddle type (Type 6) were relatively prone to pollution. In contrast, the high-pressure bottom type (Type 1) and continental high-pressure type (Type 2) were less likely to experience pollution. (2) Weather types prone to pollution were characterized by high relative humidity, low surface wind speed, low boundary layer height, and small ventilation coefficients, all of which favored local growth and accumulation of PM_2.5. Relatively prone to pollution weather types also contributed to PM_2.5 pollution when there was no significant precipitation and the surface wind speed was less than 2 m·s⁻¹. (3) Type 2, Type 3, and Type 4 was significantly influenced by external transport, with potential source regions located in the northeast and local area, widely distributed, and mainly involving short-distance transport.