Many literatures have pointed out that air pollutants increase morbidity risks of a number of diseases. It would affect the growth ofnational health expense . In this research, we are interested in estimating quantiles of air pollutants. The quantile is decomposed as a linear combination of basis functions. The quantile regression method is used to demonstrate the spatial distribution of air pollutants in different months. A subset selection method of Lasso is used to choose a suitable subset of the basis functions and estimate corresponding parameters. This method is computationally efficient for large data sets.