Using the maximum entropy (MaxEnt) algorithm and the spatial analysis module of ArcGIS, a suitable habitat prediction model for the potential geographical distribution of alpine mean in southwestern China is constructed based on the mean type map and 19 climate variables. The geographical distribution patterns of alpine mean during 1960-2000 and 2020-2050 in southwestern China are simulated. The adaptability characteristics under the future climatic change scenarios (A1B, A2 and B1) are evaluated. The results show that the applicability of the MaxEnt model for geographical distribution prediction of alpine mean in Southwest China is very high (AUC=0.93). The mean temperature in the warmest month, the mean temperature in the wettest season, and the mean temperature in the coldest month are main climatic factors limiting the distribution of alpine mean in Southwest China. The climate suitable areas for the geographical distribution of alpine mean are mainly concentrated in Tibet, Qinghai, the western part of Sichuan, and the northwest part of Yunnan, with an altitude between 4500-5500m. The proportion of totally suitable, moderately suitable, mildly suitable, and unsuitable areas in the total area is about 1:1:2:5. Under three climate change scenarios in the future, the adaptation pattern of alpine mean geographical distribution to climate change is consistent. The adaptation of alpine mean is weakened companying climate change, which is reflected by the reduction of the suitable areas of potentially geographical distribution to climate change. The altitude 5000-5500m has a strong adaptability, and the proportion of the adaptive area is the highest (about 53%). The adaptability of 3500-4500m is the weakest, and the proportion of adaptive area is the lowest (about 5%).