The in-situ monitoring of soil dehydration process is of great significance for soil and water resources management, natural disaster warning, and disaster prevention and reduction in red soil areas. The heat pulse method is commonly used for in-situ measurement of soil water and thermal characteristics; however, cracks are likely to occur when red soil loses water, it is still unclear whether the single probe heat pulse method can accurately monitor the dehydration process. This study simulated the continuous dehydration process of soil from saturation to drying by laboratory experiments, and used time-domain reflection (TDR) method as a standard to verify the accuracy of heat pulse method in monitoring soil dehydration process. In the meantime, the influence of soil crack generation on the performance of single probe heat pulse method was evaluated. The results showed that with soil dehydration, the soil thermal conductivity (K) decreased from 1.8 to 1.1 W/（m·K）, the thermal diffusion rate (α) increased from 10×10-9 to 30×10-9 m2/s, and the specific heat capacity (C) decreased from 180×106 to 40×106 J/（kg·K）. In addition, the contact thermal conductivity (H) exhibited a step-like variation characteristic with soil dehydration, which was consistent with the turning points of soil water content change rate, and the H value indicated the formation of soil cracks. Compared with the results of TDR, the MAE and RMSE values of the soil water content during the dehydration process monitored by the single probe heat pulse method were 1.66 and 1.94 cm3/cm3, respectively, with a R2 value of 0.96. The findings indicate that the single probe heat pulse method can provide reliable results for monitoring the dehydration process of red soil, and the variation of H value have an indicative significance for the generation of soil cracks. This method can provide a technical and theoretical support for the dynamic monitoring of soil wetting and drying cycle process, as well as slope and dam stability.