It is important to study the characteristics of soil water and heat to reasonably predict climate change effects and to establish an early climate warning system for the Qilian Mountains. In this study, 18 soil temperature and humidity dataloggers (HOBOU30) were used to monitor a grassland over a continuous three-year period in the western Qilian Mountains. Our aim was to explore the characteristics of soil water and heat exchange in this area and to clarify the soil water-heat interaction effect and coupling mechanism. The results showed that 1) the cold period (November to March) accounted for 42% of the year and the warm period (April to October) for 58%. As time passed, average soil temperature and water content increased monthly by 3.53℃ and 3.23%, respectively, before July and decreased monthly by 4.73℃ and 2.55%, respectively, after July. 2) Average daily soil temperature and water content were 6.45℃ and 16.26%, respectively, before 16:00, and 9.1℃ and 16.79%, respectively, after 16:00; thus, these values were 1.65℃ and 0.54% higher after 16:00 than they were before. 3) Soil temperature (P < 0.05) and water content (P > 0.05) were linearly positively correlated with altitude. 4) As soil depth increased (0-120 cm), average soil temperature increased by 0.07℃ for each soil depth, and average soil water content decreased by 0.58% for each soil depth. Soil temperature and soil depth were significantly positively linearly correlated (P < 0.05, R²=0.99), and water content and soil depth were significantly negatively linearly correlated (P < 0.05, R²=0.97). 5) Soil temperature and soil water content were negatively correlated. Therefore, the western Qilian Mountains exhibited small variation in soil temperature and large variation in water content, and both soil temperature and soil water content had a quadratic relationship with spatial variation and a linear relationship with temporal variation.