This study examined the feasibility of adopting various strategies for improving urban air quality using atmospheric visibility as an indicator. Atmospheric visibility was regularly observed twice daily with unaided eyes at two sites since November 1998 in metropolitan Kaohsiung, which was selected as the study area because of having the worst ambient air quality in Taiwan. In addition to regular observation, intensive field monitoring of atmospheric visibility and aerosol particles was conducted in January and March of 2000. Aerosol particles were sampled by using two dichotomous samplers and the analyzed for various metallic constituents, carbon content, and water-soluble ions including major anions (F-, Cl-, SO4-2, and NO3-) and cations (NH4+, K+, Na+, Ca+2, and Mg+2). The apportioning of emission sources to impairing atmospheric visibility was conducted with receptor models based on chemical mass balance (CMB) and principal component factor analysis (PCA). Receptor modeling results were then applied to establish multiple regression models for atmospheric visibility, chemical composition of aerosol particles, and apportioned emission sources. The multiple regression models were further used to determine strategies for improving urban visual air quality in metropolitan Kaohsiung.