Abstract: Based on the daily SO₂ concentration data in 6 monitoring stations of Anshan from January to May in 2007, four background values schemes were implemented to compare pollution prediction consequence by ADMS model. The different data including the intraday SO₂ concentration of monitoring stations, SO₂ concentration of monitoring stations in the day before, SO₂ concentration amplifying prediction value in proportion by ADMS model and SO2 concentration considering meteorological conditions were respectively used as background values in four schemes. The results indicate that the scheme considering meteorological conditions is the best among four schemes. Its prediction results are reasonable and its correlation coefficient and coincidence index are also the most significant.