The rising number of natural disasters and growing concerns about climate change underscore the pressing need to address regional-scale disaster risks of urban assets. One solution to this end is to evaluate optimal retrofit plans of building structures. However, such task remains challenging because of the large number of building structures with different profiles of disaster risk and retrofit cost. In addition, because of a high level of inherent uncertainties, an optimisation needs to consider uncertainties arising from disaster events and structural performance. To address the aforementioned challenges, optimization can be formulated by integrating Performance-Based Engineering (PBE) and Reliability-Based Optimization (RBO). PBE has been developed as a flexible probabilistic method that is applicable to a wide range of disasters, such as tornadoes and tsunamis. RBO is an optimization framework that minimizes decision costs while satisfying the reliability constraints. However, RBO formulations with conventional failure probability often make optimization computationally ineffective. To address this issue, this paper proposes a novel RBO method that utilizes buffered failure probability as a reliability metric. By using the metric, the proposed method formulated the optimization problem into a mixed integer linear programming, for which multiple efficient algorithms are available. We further improve computation by introducing a modified active-set technique. Thereby, the method can perform optimization over a large number of buildings. In addition, using buffered failure probability, optimization results become more insightful for risk management as it accounts for severity of failure events as well, whereas conventional failure probability deals with their likelihood only. The proposed method is demonstrated by its application to multi-disaster scenarios of earthquakes and tsunamis for a testbed of 1,000 buildings in Seaside, Oregon. To better reflect reality, the example considers a building owners' acceptance rate when being offered a subsidy for retrofit. We also perform a systematic investigation on influential factors to obtain insights for regional-level disaster risk management.