Remote sensing techniques play increasingly important role over recent decades in both problems of global climate change and frequent deterioration of the status of aquatic ecology. In this paper, Landsat ETM+ imagery was used to monitors water quality in Kunming city, southwest China. Five over 30 square kilometers lakes which are Dianchi, Fuxian, Yangzong, Qilu and Xingyun Lake were extracted and investigated. Comparing spectra of different water quality, it can be conclude the reflectance of band 2 and band 3 change dramaticlly corresponding to different water quality, which are relate to the water pollutant. Therefore, the sum of band 2 and band 3 reflectance used as a water quality index was selected to detect water quality. Firstly, the remote sensing data was atmospheric corrected and the reflectance of band 2 and band 3 was added using band math. Secondly, density slice was applied to the processed sum image using suitable data ranges and colors, and six water quality level was outputed. Finally, the result image was projected and outputed. As shown from result, the water quality of Fuxian and Yangzong lake were the best, and Dianchi and Xingyun lake were the worst, Qilu lake was in the middle. Suitable water pollution controls and provention for Kunming lakes would be needed.