Abstract: The moderate-resolution imaging spectroradiometer (MODIS) onboard NASA EOS Terra and Aqua satellites, advanced very high resolution radiometer (AVHRR) onboard NOAA series provide important aerosol measurements. MODIS provides atmosphere aerosol optical depth (AOD) product since 2000, and AVHRR also provides AOD product since 1981 but only over ocean. Developing AOD retrieval algorithm which can also obtain AOD from AVHRR over land is very important for establishing a long term AOD data record for climate studies. As 2.1 μm band is absent, an algorithm which is different from MODIS is introduced to retrieve AOD over land from AVHRR. With this method, the surface target is assumed to remain radiometrically invariant over a certain time period and some of observations are made under clear-sky background aerosol conditions. When background aerosol conditions are given, surface reflectance can be estimated by extracting the second minimum reflectance during the previous 22 days and the future 22 days. The second darkest reflectance is chosen to reduce cloud shadow contamination. After surface reflectance is selected, AOD is retrieved from a look up table (LUT) generated with the second simulation of the satellite signal in the solar spectrum (6S) radiative transfer model. The AOD over part of China (15°—45°N, 75°—135°E) from AVHRR in 2009 is obtained based on this algorithm. The distribution pattern of AOD from this work is consistent with that of MYDO04 from MODIS in North China and East China, but has some difference in Northwest China. The daily regional mean AOD from AVHRR in the Yangtze Delta (28°—36°N, 112°—122°E) agrees well with MODIS AOD with all correlation coefficients larger than 0.5 for four seasons, even up to 0.8 in winter. The correlation coefficients are 0.70 in Beijing, 0.63 in Xianghe and 0.61 in Taihu when AOD from AERONET are used to validate the AVHRR AOD retrievals. To compare temporally varying AERONET data with spatially varying AVHRR, the time match window is limited within 30 minutes and the spatial distance is limited within 0.10. The monthly variation of AOD from AVHRR in the Yangtze River Delta is consistent with that from MODIS, but the former is larger. Error sources about this retrieving algorithm are also discussed, including different satellite zenith angles in the selected period, surface reflectance, aerosol types, background AOD, calibration and sensor noise and so on. According to these results, this algorithm has the potential for deriving long-term AOD climate data record over land from AVHRR although some uncertainties still exist. Quality control and error characterization will be further investigated in the future.