Abstract: As the main component of urban green space, residential areas play an important role in improving air quality. For the first time, this research applies ENVI-met to simulate the green space layout and PM2.5 in residential areas. It makes quantitative analysis of the impacts of different grassland and tree layout on PM2.5. by selecting three typical residential areas, including Class I (R1), Class II (R2), and Class III (R3) in the city of Wuhan. The results show that the centralized layout of grassland and surrounding layout of trees can effectively reduce PM2.5 concentrations in the residential area, while the dispersed layout of grassland and centralized layout of trees are the most unfavorable to reduce PM2.5. The surrounding layout of trees has the best effect on blocking PM2.5 diffusion on the roads. Under the same green space layout pattern, R2 has the lowest mean PM2.5 concentration, and R3 has the highest mean PM2.5 concentration. The spatial distribution of PM2.5 also has some differences among the three residential areas. The research reveals the influence mechanism of green space layout on PM2.5, and provides references for the planning and design of residential green space.