Guidelines are often set at urban catchments' outfalls to avert river pollution, where stormwater is discharged into the river. Analytical probabilistic models (APMs) in conjunction with particle swarm optimization (PSO) were used to design a detention pond system at three sub-catchments of a watershed that discharge into a common point. The objective is to design multiple ponds upstream such that the pollution control target downstream is met at the minimum cost. Given the cost of purchasing land plus the cost of construction/maintenance of the ponds in the sub-catchments, the result shows that pond depths of 2.0 m in all three sub-catchments give the least total cost. A runoff control of 88, 94, and 90%, and pollution control of 59, 45, and 66% were obtained in Ponds 1, 2, and 3, respectively, while satisfying the overall watershed's pollution control target. A sensitivity analysis was conducted by varying the land costs and different performances were obtained. The APM/PSO model can search for the optimum design parameters that satisfy upstream runoff control performances and the overall pollution control target downstream. The advantage of the approach is that it can be applied to any combination of ponds in a larger watershed.