Strategic management of agricultural watersheds is essential to achieve increased production of ecosystem services (e.g., reduction of non-point source pollutants, flood peak reduction, expansion of wildlife habitat and biodiversity) and agricultural commodities. This paper presents a multi-objective optimal control model (OCM) for watershed-scale management of ecosystem services through coupling a semi-distributed hydrologic model, SWAT with a multi-objective optimization algorithm, NSGA-II. The computational model is designed to identify land uses and management practices that could result in improved production of ecosystem services and maximized net gross margin. The ecosystem service objectives considered in this study are the reduction of flood peaks and non-point source pollutants (NPS) including sediment, nitrogen and phosphorus yields whereas the economic objective is maximization of net gross margin. Based on the close relationship between these ecosystem service objectives, a hydrologic and water quality index (HWI) is introduced to indicate the level of ecosystem service generation. The HWI helps reduce the dimensionality of the overall optimization problem. Application of the OCM to the Big Creek watershed in southern Illinois indicates that the model is effective in identifying optimal tradeoffs between these conflicting objectives, i.e., improving the generation of ecosystem services and maximizing agricultural production. Thus, the OCM can be used as a decision support tool for cost-effective management of watersheds, enabling stakeholders and policy makers in identifying a suitable set of land use and management practices at a watershed scale.