Reliable model parameter estimation for stormwater runoff water quality modeling has always been a challenge, especially in ungauged areas. Insufficient local observations can increase parameter uncertainty, leading to modeling outputs of limited credibility. This paper proposes an event-based multiple pattern inverse modeling approach to estimate regional-scale parameters characterizing the buildup and washoff processes in urban watersheds. The approach explicitly accounts for system uncertainties by simultaneously using a wide range of storm events. A genetic algorithm was employed to solve the inverse models. The solution populations that minimize the error between the observed and predicted values were then filtered into distinct parameter groups using a K-means clustering algorithm. Each identified parameter group may be used to represent the buildup and washoff processes.The proposed approach was applied for the New England region (United States) using two study sites with impervious land cover: Massachusetts and New Hampshire. Stormwater monitoring data of total nitrogen (TN) and total phosphorus (TP) in small to medium size rainfall events were used to identify representative parameter sets for the region. Of the 21 identified candidate parameter combinations, 13 were found to have acceptable Root Mean Square Error for both TN and TP.The 13 parameter sets were grouped into four distinct patterns to represent different potential buildup and washoff mechanisms. The results demonstrate that the approach is capable of estimating robust sets of buildup and washoff parameters at a regional scale. These can be used in a continuous simulation model to predict TN and TP loads in ungauged areas in the region.