A contrast analysis of CAPPS1 and CAPPS2 prediction models for Guang Xi air quality was conducted, including their pollutant diffusion theories, weather forecasting models, operating systems, run time, and valid time. It is concluded that both CAPPS1 and CAPPS2 models have a correlation coefficient of over 0 44 for all kinds of pollutants, and both models are applicable to air quality forecasts in Guang Xi and have the best prediction precision on SO 2, the second on NO 2 , while PM10 prediction is relatively worse. CAPPS2 has higher correlation coefficients for all pollutants than CAPPS1 with the highest on NO 2. In contrast to CAPPS1, CAPPS2 has higher prediction precision on NO 2 and PM10 and lower prediction precision on SO 2. The prediction precision of CAPPS2 on chief pollutant is 11 percent higher than those of CAPPS1, and 9 percent higher for grade prediction.