Dam heel stress is a key indicator concerning concrete cracks and dam safety. Tension stress is generally got in a dam heel by structure mechanics or finite element methods, while seldom be monitored with instruments. The reason for the difference between simulation and monitoring results is analyzed qualitatively and quantitatively. With Xiaowan arch dam, for example, the main factors affecting simulation accuracy are studied and an improved simulation method is proposed to give more real and accurate dam heel stress. The results show that the reason mainly lie in the following: (1) different definition of dam heel between simulation and monitor; (2) limitation of monitoring methods and instruments; (3) improper simplification in simulations, such as the way to preload self-weight and water pressure, to consider thermal loads in a dam and distribution of water temperature in reservoir, and impact of basin water pressure etc. The improved simulation results are in good agreement with the monitoring results of Xiaowan arch dam. Some suggestions for improving monitoring devices and methods are given.