Abstract: Because there are not AVHRR/3 data and MODIS data before 1998, the time length of their data are not long enough to meet the needs of hydrology and meteorology. However, AVHRR/2 data can extend to 30 years ago. For AVHRR/2 data don't have the data of 1.6 μm waveband, which AVHRR/3 data and MODIS data have, the existing subpixel snow mapping algorithm of moderate resolution data cannot be used to implement the subpixel snow mapping of AVHRR/2 data. Therefore, in order to satisfy the needs of hydrological and meteorological modeling for the long-term data of snow covered area, an algorithm of subpixel snow mapping was established using AVHRR/2 10-day composited data of Qinghai-Tibet Plateau in the work. Moreover, linear spectral mixture model and multiple endmember spectral mixture analysis were introduced into the algorithm. According to the algorithm, the percent data of subpixel snow covered area can be obtained from AVHRR/2 data, and the subpixel snow covered area data can be extended to more than 30 years.For obtaining the required endmembers of ground objects, the method of selecting endmembers was found, in light of which snow endmembers and non-snow endmembers were selected automatically from AVHRR/2 images using the data of NDVI, channel 1 and channel 2 as the indices of selecting the endmembers, and the obtained snow and non-snow endmembers were saved in respective databases. In the work, this algorithm was validated against TM data, and the root-mean-square errors of the validation result are close to 0.1. For such a mountainous region as Qinghai-Tibet Plateau, this algorithm is very reliable and feasible.Because the used AVHRR/2 data are the 10-day composed data, the daily subpixel snow covered area cannot be obtained. In future, it is necessary to implement the pre-processing of AVHRR/2 data, and to obtain usable daily AVHRR/2 data. Then, after the obtained daily AVHRR/2 data were mapped at subpixel scale, the daily subpixel snow area data from AVHRR/2 data will be obtained.