Operational air quality forecasts of ground-level ozone and particle pollution to protect public health have become routine in the United States. Various tools such as regression, photochemical models, and phenomenological rules have been developed to provide guidance for enhancing the accuracy of air quality forecasts. However, the accuracy of these forecast tools has yet to be uniformly and consistently evaluated. This paper provides an in-depth and objective assessment of human forecasts, regression equations, and the NOAA-EPA and Canadian CHRONOS forecast models for six selected East Coast cities. It also discusses the challenges faced by air quality forecasters in making more accurate and specific forecasts as well as the future direction and advancement of air quality forecasts nationwide.