To improve convenience and safety for pedestrians, signalized two-stage, midblock pedestrian crosswalks are increasingly being installed in highly populated areas in developing countries such as China. This paper presents a multiobjective optimization model and its solution algorithm for optimal control of a two-stage, midblock crosswalk on a street with both vehicular and pedestrian traffic. The proposed model aims to produce the optimal signal timings at the crosswalk to accommodate both traffic modes and to adjust the offsets of the pedestrian signals for the two stages concurrently to minimize pedestrian delays and relieve congestion at a central refuge island, the safe area for pede.strians to stop. The proposed model and algorithm have three distinguishing features: (a) they explicitly model pedestrian delays at the two-stage controlled crosswalk, including delays at both the curbside and the central refuge island; (b) they model the number of pedestrians waiting on the central refuge island according to the change in signal timing; and (c) they have application in a multiobjective optimization approach to study the effectiveness of midblock crosswalk control under conditions in which the priorities between vehicular and pedestrian traffic differ. A heuristic based on nondominated sorting genetic algorithm II was designed to solve the model and generate the Pareto solution set for signal timings. The results of the case study showed that the proposed model would help traffic practitioners, researchers, and authorities properly design signal timing plans and central refuge islands for two-stage midblock pedestrian crosswalks.