Although there is a great deal of information concerning responses to increases in atmospheric CO₂ at the tissue and plant levels, there are substantially fewer studies that have investigated ecosystem-level responses in the context of integrated carbon, water, and nutrient cycles. Because our understanding of ecosystem responses to elevated CO₂ is incomplete, modeling is a tool that can be used to investigate the role of plant and soil interactions in the response of terrestrial ecosystems to elevated CO₂. In this study, we analyze the responses of net primary production (NPP) to doubled CO₂ from 355 to 710 ppmv among three biogeochemistry models in the Vegetation/Ecosystem Modeling and Analysis Project (VEMAP): BIOME-BGC (BioGeochemical Cycles), Century, and the Terrestrial Ecosystem Model (TEM). For the conterminous United States, doubled atmospheric CO₂ causes NPP to increase by 5% in Century, 8% in TEM, and 11% in BIOME-BGC. Multiple regression analyses between the NPP response to doubled CO₂ and the mean annual temperature and annual precipitation of biomes or grid cells indicate that there are negative relationships between precipitation and the response of NPP to doubled CO₂ for all three models. In contrast, there are different relationships between temperature and the response of NPP to doubled CO₂ for the three models: there is a negative relationship in the responses of BIOME-BGC, no relationship in the responses of Century, and a positive relationship in the responses of TEM. In BIOME-BGC, the NPP response to doubled CO₂ is controlled by the change in transpiration associated with reduced leaf conductance to water vapor. This change affects soil water, then leaf area development and, finally, NPP. In Century, the response of NPP to doubled CO₂ is controlled by changes in decomposition rates associated with increased soil moisture that results from reduced evapotranspiration. This change affects nitrogen availability for plants, which influences NPP. In TEM, the NPP response to doubled CO₂ is controlled by increased carboxylation which is modified by canopy conductance and the degree to which nitrogen constraints cause down-regulation of photosynthesis. The implementation of these different mechanisms has consequences for the spatial pattern of NPP responses, and represents, in part, conceptual uncertainty about controls over NPP responses. Progress in reducing these uncertainties requires research focused at the ecosystem level to understand how interactions between the carbon, nitrogen, and water cycles influence the response of NPP to elevated atmospheric CO₂.