Abstract: The study on separation of meteorological information and emission information of pollution sources in air pollutant time series is of some important significance to evaluation of air pollution control measures. Based on the daily air pollution index (API) and simultaneous surface meteorological observation data during 2001-2010 in Beijing, Tianjin and Shijiazhuang, the Kolmogorov-Zurbenko (KZ) wave filter method was used to break the API time series into short-term, seasonal and long-term components, and then the stepwise regression was used to set up API baseline and short-term components separately and establish linear regression models for meteorological variables of corresponding scales. The residual differences between the baseline and short-term components were KZ filtered and series reestablished. And eventually an API time series free of meteorological impacts was worked out. Results of the study show that the long-term component of API shows a downward trend in all the three cities from 2002-2009, which may be attributed to changes in emission with the influence of meteorology removed, namely, meteorological adjustment. Comparison of the meteorologicaly adjusted long-term trends with unadjusted ones show that unfavorable meteorological conditions often lead to increase by 1-13 in the long-term API in the three cities, whereas favorable meteorological conditions account for reduction by 2-6 in the component.