This thesis contributes to the field of spare parts inventory management by developing robust optimization models that effectively handle high demand uncertainty across three key supply chain settings: local warehouses with lost sales, local warehouses with emergency shipments, and central warehouses with backorders.nnFor local warehouses with lost sales, the author introduces an Adjustable Robust Optimization (ARO) model and proposes efficient algorithms like IPDO, ConGA, and LES to solve large-scale problems. These methods balance solution quality and computational speed, demonstrated through a case study at ASML involving 710 components.nnIn local warehouses with emergency shipments, the thesis presents new ARO models and algorithms (ISP and ConGAP) that incorporate limited historical data and initial failure rates to construct uncertainty sets. The ASML case study shows that these models significantly reduce waiting times and potential production losses.nnFor central warehouses with backorders, a novel continuous review ARO model is developed. A three-step solution approach is used to handle varying lead times and optimize stock levels efficiently. Again, the ASML case study confirms the model’s cost-effectiveness, especially under strict service level requirements.nnOverall, the thesis advances robust optimization techniques, scalable algorithms, and uncertainty modeling, offering practical and high-performing solutions for spare parts inventory control. Future research could explore lead-time uncertainty, nonlinear costs, and machine learning integration.