The utilization of random finite element method (RFEM) to make geotechnical prediction has gained popularity in recent decades. These stochastic analyses, which accounts for the spatial variability of soil, can more precisely reflect soil condition in the natural environment. This study discusses the required domain size for footing bearing capacity analysis with RFEM, which can be applied to many infrastructure designs such as road pavement, airport runway, shallow foundation of buildings and so on. Since the infrastructure sits on an unbounded ground, expanding the domain size in RFEM can enhance the accuracy of result. However, it is found that after a critical size, any larger domain size would not contribute to a significant improved accuracy. The appropriate domain size proposed in this study is not limited to a narrow set of parameters. Instead, it is applicable across a wide range of situations, including different combinations of material characteristics. The robustness of our finding is verified by many numerical simulations, which shows the reliability of the finding in diverse geotechnical scenarios.