Rhodobacter sphaeroides is a versatile purple non-sulfur bacteria that can be used in resource recovery to produce hydrogen, polyhydroxyalkanoates, single-cell protein and other value-added products. We developed a genome-scale metabolic model representing the primary metabolism of a denitrifying strain R. sphaeroides ATCC 17025. This metabolic model predicted growth, production of hydrogen and polyhydroxybutyrate, and was in better agreement with literaturebased experimental values as compared to predictions from existing models for R. sphaeroides. We evaluated the growth, production rates and product distribution under different metabolisms and using different organic substrates. In addition, it allowed to assess the metabolic mechanisms governing the preferential generation of different products. This model can be a valuable tool to optimize the generation of value-added products.