Precipitation phase has a direct influence on the surface energy balance and hydrological processes. It is important and necessary to separate the snowfall from precipitation and quantify the impact of climate change on the snowfall for the improvement of snowmelt runoff simulation and the management of the water resources in the alpine area. Based on the relation between snowfall proportion and daily temperature of six meteorological stations before 1979 in upper reaches of Nujiang River Basin, a separation scheme with exponential function structure was applied to estimate the snowfall,which was validated by the precipitation phase data before 1979. Furthermore, the snowfall was estimated and analyzed by the separation scheme from 1980 to 2016, and the impact of temperature and precipitation change on snowfall evolution was explored. The results show that:(1) the separation scheme performs well in snowfall estimation with Pearson correlation coefficient above 0.86 and relative BIAS under 5% in the six stations. (2) Temperatures after 1979 show a significant increasing trend in the six stations,while the annual precipitation and snowfall demonstrate insignificant increasing and decreasing trends overall the basin, respectively. (3) Increase of temperature promotes the decrease of snowfall, and the influence intensity of temperature on the snowfall gradually increases from the southeast to the northwest of basin. Moreover, the precipitation probability increases in the low temperature condition under the climate change,which significantly promotes the snowfall increase in upper reaches of Nujiang River Basin. However, the significant increase of temperature has a dominant effect on the snowfall change after 1979,and the overall snowfall shows a decreasing trend in upper reaches of Nujiang River Basin.