Particulate matter with an aerodynamic diameter of 2.5 μm or less (PM2.5) is a major concern in many areas. Numerous epidemiological studies have identified the adverse health effects from exposures to PM2.5. From the perspective of reducing exposures and protecting public health, this presentation will cover two approaches of identifying PM2.5 sources and their associations with health outcomes. (1) Receptor modeling: The concepts of source apportionment will be introduced together with examples of PM2.5 sources identified in Taiwan using the Positive Matrix Factorization (PMF) model. The source-specific health analysis results also will be discussed. (2) Land use regression: Through collecting air monitoring data at multiple locations and retrieving predictor variables based on the geographic information system (GIS) and regression models, sources affecting PM2.5 distribution can be identified properly. Examples of land use regression researches and their application in epidemiological studies will be presented.