This study presents a statistical design method in developing magnesium phosphate cement (MPC) -based material for rapid-repair. Specifically, an optimization model considering the relationship between variables (the water-to-binder ratio, the magnesia-to-phosphate ratio, and the borax content) and responses (the setting time and the early-age compressive strength) has been established via response surface methodology (RSM). After the accuracy evaluation, MPC performance optimization statistical model has been developed. The results show that the water-to-binder mass ratio(w/b) is the most critical factor aiming to attain a higher early-age compressive strength while maintaining a suitable setting time, followed by the borax-to-magnesia mass ratio (B/M) and magnesia-to-phosphate molar ratio (M/P). The results shown that the final setting time gradually prolongs with the increase of w/b and B/M, while the decrease of M/P. With regards to the early age compressive strength, where the strength increases at first then decreases with the increase of w/b and M/P, while generally reduces with the addition of borax. Moreover, it can be concluded form the statistical model that the value of the optimal M/P depends on the level of w/b. (The interaction may be ignored due to the nature of typical performance-based design, thus, RSM contributes a more accurate theoretical solution). After that, two optimized theoretical solution results have been developed. After experimental verification, the optimized results are also compared with results developed by other workers, where the improved properties further emphasize the advantages of RSM in design MPC. This study demonstrates the potential of RSM in developing MPC in patch repair and maintenance works of municipal engineering as a greater scale.