Abstract: This paper describes the introduction of a photo-chemistry model based on a highly parameterized chemistry mechanism into the nonstatic atmospheric advection-dispersion box-model to practise ozone prediction in urban boundary layer. Genetic algorithm is employed to inverse the VOC (volatile organic compounds) concentration as well as NO_X source strength since the ambient pollutant concentration and emission data for urban area are insufficient. this work gives the birth to a new version of the City Air Pollution Prediction System (CAPPS) with the atmospheric chemical processes included. Based on regular sounding data and the pollutant concentration data collected at Beijing Dance Academy, testing predictions of ozone and NO_X are performed using the new single-box model. The results indicate that the new model is proved to be an effective tool for urban area ozone prediction.