Abstract: Atmospheric PM_2.5 pollution is one of the major ecological and environmental problems in China. It is important to investigate the spatial and temporal evolution characteristics of PM_2.5 pollution and its population exposure risk, and to reveal the influence of landscape pattern on atmospheric PM_2.5 pollution, in order to reveal the atmospheric pollution pattern and then improve the human living environment. Based on PM_2.5 and population grid data, exposure risk model and trend analysis model were used to explore the spatialotemporal evolution characteristics of PM_2.5 population exposure risk in southwest China from 2000 to 2020. In this study, the landscape pattern index was measured based on the best moving window using 3-period land use data, and the correlation coefficient method was used to reveal the relationship between the landscape pattern and PM_2.5 and its population exposure risk in the study area. The results show that:(1) The overall annual average PM_2.5 concentration in the southwest region and provinces showed a slow increase and then a rapid decrease during the study period, and there were significant east-west differences in its spatial distribution.(2) The spatial distribution pattern of PM_2.5 population exposure risk was relatively similar in all years, and the high, higher and very high risk areas were mostly distributed in urban areas with concentrated population. In terms of spatial variation, the areas with very significant decrease in PM_2.5 population exposure risk were mostly found in the central urban areas of prefecture-level cities, and the areas with very significant increase were mostly found in the urban fringe and small and medium-sized towns.(3) Among the landscape pattern characteristics in Southwest China, the proportion of landscape has a significant effect on PM_2.5 concentration, but the edge density, shape index and average patch area also play different roles in different landscapes. Within the overall landscape pattern characteristics, increases in landscape fragmentation, patch shape complexity and sprawl would lead to higher annual average PM_2.5 concentrations and their enhanced risk with population exposure.