Mission engineering is a growing field with many practical opportunities and challenges. The goal of mission engineering is to increase system effectiveness, reduce life cycle costs, and aid in communicating system capabilities to key stakeholders. Optimizing system designs for their mission context is important to achieving these goals. However, system optimization is generally done using multiple key performance indicators (KPIs), which are not always directly representative of, nor easily translatable to, mission success. This paper introduces, motivates, and proposes a new approach for performing mission-level optimization (MLO), where the objective is to design systems that maximize the probability of mission success over the system life cycle. This builds on previous literature related to mission engineering, modeling, and analysis, as well as optimization under uncertainty. MLO problems are unique in their high levels of design, operational, and environmental uncertainty, as well as the single binary objective representing mission success or failure. By optimizing for mission success, designers can account for large numbers of KPIs and external factors when determining the best possible system design.