Aerosol Optical Depth (AOD) is a vitally important physical parameter of aerosol which can be retrieved from satellite remote sensing data. The traditional algorithm such as MODIS aerosol algorithm is only appropriate for the dark pixels which have low reflectance. If the surface reflectance is high, the algorithm is ineffective. In this paper we proposed an advanced quantitative retrieval algorithm for AOD over land from MODIS data. In our new algorithm, MODIS L1B reflectance values are corrected for water vapor, ozone and carbon dioxide before used for further AOT calculation. Then we applied our algorithm to retrieve AOD of Beijing area in May and June, 2007. Preliminary validation result comparing with AERONET measured data shows good accuracy and promising potential with relative errors reduced from around 10% to below 10% (around 5%). Further research work is ongoing.