In this paper, the design of the air pollution forecast and early warning system of Karaj metropolis is presented. This system is one of the first systems in Iran in this field. For this purpose, three models including the emission inventory model, meteorological model, and photochemical air quality model were coupled. Using provided codes and scripts the modeling system runs automatically every day and 48 hours forecast of air pollutant concentrations (including carbon monoxide, nitrogen dioxide, particulate matter, sulfur dioxide, and ozone) and meteorological fields are sent to the web-based early warning system. In the system, the forecast of air quality index of different regions of Karaj is displayed separately by regions and on an hourly and daily basis. The results of the meteorological model for temperature and wind speed have a correlation of 0.96 and 0.78 with the observations, respectively. The results of the air quality model for the ozone pollutants differ by an average of 20 ppb in the worst case with the observation. The designed web-based system receives the results and presents them in the form of maps and reports.