A new technique for forecasting meteorological conditions of air pollution is developed by the analysis based on Air Quality Index (AQI) of PM2.5 from 2013 to 2016. According to the correlation analysis between AQI and meteorological factors, the IBAM (Index Between Air pollution and Meteorology) is built by four factors (specific humidity, surface wind velocity, surface pressure change in 24 hours, and the difference of total temperature between two levels in the lower troposphere) from 2013 to 2016 in Chongqing. Then the past IBAM is attained by use of the EC numerical forecast products from 1 April 2013 to 31 〖JP2〗December 2016, and the prediction model of meteorological conditions is established by introducing the concepts of extreme weather events and Kmeans cluster analysis. The fitting curve equation between IBAM and AQI of 1day lagging can be used to calculate the forecast value of AQI; then according to the ranking standards of air quality, the forecasting accuracy of air pollution grades in 72h prediction reaches about 70% by a test with the samples of nearly two years (from 1 January 2017 to 1 September 2018). Through the error analysis of two air pollution events, the results show that the prediction of air pollution grades is relatively well when the meteorological condition is the major factor impacting the spread of the pollutants in atmosphere or the change of air pollution sources is not obvious (local accumulation as main air pollution mode). However, the prediction errors increase evidently when the change of air pollution sources is relatively remarkable because of upstream transportation. This forecast technique is applied in the realtime prediction services in the Chongqing Meteorological Observatory, which has important reference value for preventing heavy air pollution events and improving air quality in Chongqing.〖JP〗