For the first time, we combine concepts of theoretical food web modeling, the metabolic theory of ecology, and ecological stoichiometry with the use of functional trait databases to predict the coexistence of invertebrate taxa in streams. We developed a mechanistic model that describes growth, death, and respiration of different taxa dependent on various environmental influence factors to estimate survival or extinction. Parameter and input uncertainty is propagated to model results. Such a model is needed to test our current quantitative understanding of ecosystem structure and function and to predict effects of anthropogenic impacts and restoration efforts. The model was tested using macroinvertebrate monitoring data from a catchment of the Swiss Plateau. Even without fitting model parameters, the model is able to represent key patterns of the coexistence structure of invertebrates at sites varying in external conditions (litter input, shading, water quality). This confirms the suitability of the model concept. More comprehensive testing and resulting model adaptations will further increase the predictive accuracy of the model.