Many coastal seas worldwide are affected by human impacts such as eutrophication, causing, inter alia, oxygen depletion and extensive areas of hypoxia. Depending on the region, global warming may reinforce these environmental changes by reducing air–sea oxygen fluxes, intensifying internal nutrient cycling and increasing river-borne nutrient loads. The development of appropriate management plans to more effectively protect the marine environment requires projections of future marine ecosystem states. However, projections with regional climate models commonly suffer from shortcomings in the driving global General Circulation Models (GCMs). The differing sensitivities of GCMs to increased greenhouse gas emissions impact regional projections considerably. In this study, we focused on one of the most threatened coastal seas, the Baltic Sea, and estimated uncertainties in projections due to GCM deficiencies relative to uncertainties caused by future greenhouse gas emissions and nutrient load scenarios. To address the latter, transient simulations of the period 1975–2098 were performed using the initial conditions from an earlier reconstruction with the same Baltic Sea model (starting in 1850). To estimate the impacts of GCM deficiencies, dynamical downscaling experiments with four driving global models were carried out for two greenhouse gas emission scenarios, RCP4.5 and 8.5, and for three nutrient load scenarios covering the plausible range between low and high loads. The results of primary production, nitrogen fixation, and hypoxic areas show that uncertainties caused by the various nutrient load scenarios are greater than the uncertainties due to global model deficiencies and future greenhouse gas emissions. In all scenario simulations, a proposed nutrient load abatement strategy, i.e., the Baltic Sea Action Plan, will lead to a significant improvement in the overall environmental state. However, the projections cannot provide detailed information on the timing and the reductions of future hypoxic areas due to uncertainties in salinity projections caused by uncertainties in projections of the regional water cycle and of the global mean sea level rise.