Emergency rescue facility is an indispensable component of urban system and important for public safety. Due to the special characteristics and complexity of urban environmental accidents, locations of emergency rescue facilities should be optimized, and the decision criteria should be comprehensive, taking into account not only the spatial relations of these facilities but also the potential hazard of environmental accidents. This study proposes an innovative methodological framework to optimize the locations of emergency rescue facilities in the contexts of large-scale urban environmental accidents. Risk-based approach is involved to delineate the potential hazard of environmental accidents to the adjacent urban areas, and a three-objective decision optimization model is constructed. Moreover, an appropriate spatial representation and encoding strategy is designed and coupled with the non-dominated sorting genetic algorithm-II for model solving. A case study is presented to demonstrate the methodology, and it provides evidence that the involvement of risk mapping into the model construction has significant effect on the optimization result. The findings show that the strategy proposed in this study has the potential to be a useful decision support tool for urban planning with respect to public safety.