<p>Deep excavations require a supporting system to maintain stability and prevent excessive deformation, but the inherent spatial variability of soil properties complicates the balance between system safety and construction cost. To address this challenge, this study proposes a robust design optimization framework that considers sandy soil spatial variability to improve supported excavation designs. In this framework, the random field theory is adopted to simulate the spatial variability of sandy soil properties along both the vertical and horizontal directions, incorporating multiple key parameters. The construction cost and design robustness based on various system responses of supported excavations are selected as design objectives, constrained by different safety requirements. The Pareto front is constructed in this proposed framework and the knee point is identified as the most preferred design based on the gain–sacrifice relationship on the Pareto front. The proposed framework is demonstrated through a case study, where the effects of different levels of soil coefficient of variation (COV) and safety requirements represented by target failure probability on the optimization results are evaluated. The results indicate that higher soil variability and more stringent target failure probability requirements limit the number of non-dominated optimal designs on the Pareto front, which can have an impact on the selection of the most preferred design. The study highlights the importance of incorporating the soil spatial variability into the robust design optimization and provides valuable insights into the impacts of soil spatial variability on the robust design optimization of supported excavation systems.</p>