Abstract: Abstract: As a significant part of water resources in the ecosystem cycle, soil moisture plays an extremely important role in the process of matter and energy exchange between the surface and the atmosphere. But it is very difficult to accurately measure large scale soil moisture. Cosmic-ray neutron method is a promising way to measure soil moisture for intermediate spatial scales. This method bridges the scale gap between point measurements of soil moisture and remote sensing, making significant contribution to the measurement of soil moisture within a regional level. In 2008, Zreda et al. introduced a method to measure average soil water content over a larger area with a cosmic-ray neutron sensor (CRS). The footprint of a CRS covers a circle with an approximate radius of 300 m and the effective measurement depth varies between 10 and 70 cm below the soil surface. Desilets et al. proposed an equation with three constant shape parameters (a0, a1, a2) and one calibration parameter (N0), which needs to be calibrated with soil moisture values determined by oven-drying method from field soil samples in 2010. Franz et al. (2013) developed a universal calibration function for determination of soil moisture with cosmic-ray neutrons that takes into account three influencing factors including pressure, incoming neutron flux and water vapor in the air. Meanwhile, other scholars have explored other factors that influence the soil moisture content, including the depth and range of detection, the lattice water, soil organic carbon and vegetation biomass. In this study, experiments were conducted in China Agricultural University Experiment Station with commercial cosmic-ray neutron sensors (CRS1000) and self-assembly 3He tube neutron moisture detector (3He tube). In order to verify the accuracy and stability of these two instruments, we compared the soil moisture content data with the result from oven-drying method. Meanwhile, we made a comparison between the sensitivity of these two instruments for the response of precipitation event. Before using the two instruments, we needed to calibrate the N0 from neutron conversion equation. And the calibration was based on soil moisture content derived directly from soil samples taken within the footprint of the sensor. After one correction, N0 had certain stability over a long time. The intensity of the incoming neutron was one factor that affected the neutron count of the instruments, and it was necessary to calibrate the intensity of the incoming neutron when the solar activity was intense. Oven-drying method was used to measure the average moisture content within the footprint of the sensors. Then we used the average moisture content data to verify the stability of the two kinds of measurement. Comparing the soil moisture content data obtained from oven-drying method, it can be calculated that the root mean square error of the two instruments were respectively 0.036 and 0.015 cm3/cm3. It showed that the measurement results of two instruments were more accurate. The original neutron counts per hour of 3He tube were about 10 times more than CRS1000, so we can conclude that the former was far more accurate than the latter. After corrected, soil moisture measurement results of CRS1000 and 3He tube were more consistent. When precipitation event occurred, the change in 3He tube was even more pronounced than CRS1000, showing that 3He tube was more sensitive than CRS1000. At the same time, the cost of self-assembly 3He tube neutron moisture detector was about half of commercial cosmic-ray neutron sensors. So 3He tube will have a better application prospect.