The urbanization of an undeveloped area often shortens outflow travel time and increases peak discharge from the basin, thereby increasing the downstream flood frequency. Stormwater detention ponds are the most common measure to maintain the outflow from the post-developed basin to a flow similar to that under the pre-developed condition. The design criteria of stormwater detention ponds are to minimize construction cost while achieving the flood control purpose. The tedious and time-consuming, trial-and-error method is commonly used to determine the optimal size and location of the pond and outlet structure for a design storm period. In this study, a stochastic search algorithm, a Genetic Algorithm (GA), is used to optimize the detention pond design. The decision variables are the pond storage, and the pipe diameters and number of pipes for the service outlet. The flood control objective considered in this study is that the peak discharge in the post-developed condition does not exceed that under the pre-developed condition and that the maximum water level in the pond during the flood remains below the allowable water level. The proposed optimization algorithm method was applied to the real design of two detention ponds in South Korea, where it generated better design options comprising smaller pond storage and smaller outlet standpipe dimensions than those of the traditional trial-and-error method, and in a much shorter computational time. Therefore, the stochastic search algorithm, GA, can be successfully applied in the design of a stormwater detention basin to improve accuracy and convenience. The engineers can accordingly assess the development plan in terms of the potential basin disaster more efficiently than is possible when using the tedious computation method.