Abstract: Temperature control and crack prevention of roller-compacted concrete（RCC） dams are the key technical problems in the construction of projects. The material parameters of RCC dams have significant spatial variability due to the large pouring concrete surface, uneven rolling quality, and heterogeneity of concrete. The deterministic calculation is adopted in the traditional temperature and stress analysis of concrete dams,which ignores the influence of parameter spatial variability, resulting in unreasonable simulation of temperature and stress field. Based on the random field theory and the center point discrete method, a finite element simulation method of spatial variability of concrete thermal parameters is proposed in this paper. The case study of TJ dam shows that the average effect of the maximum temperature field and maximum tensile stress field considering the spatial variability of thermal parameters is basically consistent with the deterministic results.The spatial variability of thermal parameters significantly affects the maximum temperature and maximum stress of the dam concrete structure. The temperature and stress at some positions of structures in the random temperature and stress field exceed the allowable value, which means that the temperature control measures proposed based on the traditional deterministic temperature stress analysis results may be dangerous. The random temperature and stress analysis method proposed in this paper provides methodological support for the formulation of temperature control measures of gravity dams during the construction period.