Observation and analysis of PM_2.5 particles on a leaf surface can help us explore the correlation between PM_2.5 particles and environmental factors and the response of stomata to particulate pollution in the air. This not only provides an important scientific reference for elucidating the mechanism of PM_2.5 retention and absorption by leaves, but is also helpful for reasonably selecting and optimizing urban greening tree species to reduce particulate matter pollution and improve the atmospheric environment in the city. In Beijing, National Olympic Forest Park and Xizhimen traffic hub are two representative sites for monitoring the environment. In June and September 2013, we measured the atmospheric PM_2.5 level of these monitoring sites using a real-time PM_2.5 monitoring instrument, and found that the average atmospheric PM_2.5 level in Xizhimen traffic hub was about 1.5 times higher than that in the National Olympic Forest Park. In this study, PM_2.5 particles retained on the leaf surface of Chinese white poplar (Populus tomentosa Carr.) from these two sites were identified, counted, and analyzed by using an environmental scanning electron microscope (ESEM) and an X-ray energy dispersive spectrometer (EDS). At the same time, the adaptive response of leaf stomata to the PM_2.5 pollution was studied. These results showed that the numbers of PM_2.5 particles on both the adaxial and abaxial surfaces of leaf samples at the Xizhimen traffic hub were more than those in the National Olympic Forest Park in both summer and autumn, indicating that the number of PM_2.5 particles adsorbed by leaves is largely affected by the atmospheric PM_2.5 concentration. The adaxial surface of leaves is the main area of PM_2.5 retention. The number of PM_2.5 particles on the autumn samples was more than that on the summer samples. The PM_2.5 particles on the leaves from the National Olympic Forest Park were mainly composed of aluminosilicate particles and quartz grains, and these particles are mainly from soil dust and mineral particles. However, the composition of PM_2.5 particles from the Xizhimen traffic hub samples was more complex. The spectral peaks of Cu, Na, K, Cl, and other elements were found in the energy spectra in most aluminosilicate particles. These elements are mainly from industrial emissions. The sulfur content of PM_2.5 from the Xizhimen traffic hub samples was higher than that from the National Olympic Forest Park samples in both summer and autumn. Moreover, the sulfur content was higher in summer than in autumn. Our results indicate that the amount and composition of particulate matter are affected by environmental factors to a great extent. It was also observed that a few PM_2.5 particles are present in the stomata of P. tomentosa leaves. Also, the morphological characteristics of stomata varied at different atmospheric PM_2.5 levels. The length, width, area, and density of stomata of the Xizhimen traffic hub samples were smaller than those of the National Olympic Forest park samples due to more serious PM_2.5 pollution. Additionally, the secondary electron image and the back-scattering image were combined to better locate, observe, and analyze PM_2.5 particles on the leaf surface of P. tomentosa.