PM_2.5 data from the national air quality monitoring network in China suffered from significant inconsistency and inhomogeneity issues. To create a coherent PM_2.5 concentration dataset to advance our understanding of haze pollution and its impact on weather and climate, we homogenized this PM_2.5 dataset between 2015 and 2019 after filling in the data gaps. The homogenized PM_2.5 data is found to better characterize the variation of aerosol in space and time compared to the original dataset.