Abstract: The water vapor changes in arid areas could affect the structure and evolution of water resource systems in their surrounding areas. Based on the precipitable atmospheric water vapor (GPS-PWV) of 4 ground-based GPS stations, the observation data of 2 sounding stations and the hourly surface pressure water vapor data of 108 meteorological observation stations on the north slope of the Middle Kunlun Mountains from January 2020 to December 2022, this study established the atmospheric water vapor content and surface water vapor pressure (W-e) model suitable for the north slope of the Middle Kunlun Mountains using the unary linear fitting method. The results of water vapor content calculated by this model were verified. Then we analyzed the distribution characteristics of atmospheric water vapor content in the western section, the middle section, and the eastern section of the study area, as well as the relationship between the beginning time of precipitation and the W-PWV peak value. The results show that: (1) The annual mean W-PWV is largest in the western section of the study area, followed by the middle section, and the smallest in the eastern section which located in the southern edge of the desert. The W-PWV of the stations with altitude greater than 1 500 m gradually decrease with altitude increasing. The average W-PWV of each meteorological observation station in summer is about twice than that in spring and autumn. (2) The monthly variation of W-PWV shows a unimodal distribution characteristic. The W-PWV of the stations with an altitude higher than 1 300 m but lower than 1 500 m reached its peak in July and August, while that of the other stations reached its peak in August. The W-PWV of stations with an altitude below 2 000 m and above 2 000 m maintained a high value at night and during the day, respectively, which may be related to the thermal difference between mountain and basin from daytime to nighttime. (3) There is a good correspondence between the W-PWV calculated by W-e model and the beginning time of precipitation. Before precipitation, the W-PWV of each station is jumped varying degrees, and the peak value of W-PWV within 1-2 h before the precipitation is more than 1.5 times of the monthly average value of W-PWV.