Abstract: In order to make forecast for urban O₃ pollution in China,a numerical forecasting model system on O₃ pollution is developed recently.The system consists of models on physical and biological processes (including pollution source model,parameterization model on underlying surface,meso-α meteorological model,meso-β meteorological model,parameterization model on turbulent statistics,and dry and wet deposition model) and high resolution O₃ chemical model (HROM).The system was used to do O₃ practical forecasting experiment for 24 h in Chongqing,Guangzhou and Jinan cities.The results showed that there is good consistency between forecasted and measured O₃ mass concentrations and the average forecasting accuracy of API can exceed 85%;there are obvious diurnal variation,heterogeneity of spatial distributions,and regional difference for urban O₃ mass concentrations;NO₂ is an important precursory pollutant of O₃,and there is a negative correlation well between NO₂ and O₃.