Abstract: Soil moisture is one of the most important indices for agricultural drought monitoring and water resources management. Remote sensing is a critical technology for monitoring spatial and temporal variations in soil water content. The thermal inertia method, which is a thermal infra red (IR) technology, has demonstrated advantages in monitoring soil water condition. Among the several models for computing soil thermal inertia by remote sensing, ascertaining the conditions for monitoring soil water content by thermal inertia remains a major obstacle. This paper proposed an improved model for calculating Apparent Thermal Inertia (ATI). In the first step, a new soil ATI model with improved algorithms for simulating net radiation was developed. Then a strict control ground experiment was conducted to test the proposed model. A total of 10 experimental plots with different vegetation covers and soil water contents were set up at the Luancheng Agro-Ecosystem Experimental Station of Chinese Academy of Sciences. The vegetation covers were fully representative by NDVI (normalized difference vegetation index). The actual measured land surface temperature, NDVI, albedo, soil water content, solar radiation and long-wave atmospheric radiation were used to compute ATI under different land cover and soil water conditions. Then correlation and regression analyses were finally done to relate ATI and soil water content. The results indicated that the proposed thermal inertia model reliably monitored the soil water condition, especially in low vegetation cover areas. For low vegetation cover (NDVI < 0.35), the coefficient of determination between ATI and soil volumetric water content was > 0.7. The proposed thermal inertia method was invalid for NDVI > 0.35 and the corresponding coefficient of determination was < 0.2. NDVI that was the equivalent of 0.35 could be critical for determining the applicability of the proposed model in monitoring soil water conditions. This was because temperature dynamics (the most critical criteria for calculating ATI) for bare and vegetated lands were different. However, the proposed model was not only simple, but it carries distinct physical meaning and easy-to-use interfaces. The experiment suggested that the model was applicable in reliably monitoring soil water conditions.