Abstract: Based on the daily maximum, minimum and mean temperature data of 51 stations in Xinjiang Uygur Autonomous Region from 1961 to 2016, the spatial and temporal characteristics of 16 extreme temperature indices in Xinjiang are studied. The methods include linear trend estimation, rescaled range analysis, spatial analysis and Pearson correlation analysis. The results are as follows: (1) there is a good correlation between extreme temperature and annual average temperature in Xinjiang. This indicates that the extreme temperature changes in the region can be used as an indicator of climate change; (2) the extreme temperature indices have a significant increasing trend from both spatial and temporal point; (3) among sixteen indices, the increasing trend of these indices (TN90p, TN10p, FD0 and GSL) is larger than that of the others during the last 56 years. The warming trend in these areas (Shisanjianfang, Turpan, Barkol and Hami) is the largest from spatial point. (3) the warming magnitude of the cold (nights) indices is larger than that of the warm (days) indices, which indicates that the cold (nights) indices are more sensitive to global warming; (4) the long-term persistence of the changes in the extreme temperature indices indicates that the warming trend will continue in the future; (5) the extreme temperature indices in Xinjiang are related more closely to longitude and elevation than to latitude.