► Data from a FACE experiment are compared with an ecohydrological model. ► Modeled differences between CO2 scenarios are within the uncertainty of observations. ► Changes between CO2 scenarios in terms of water and energy fluxes are small. ► Testing the carbon allocation is hampered by current accuracy of field data. Projections of the future carbon and water cycles rely on knowledge on how forests will respond to rising atmospheric CO2. Experiments with elevated CO2 are logistically challenging and carbon pools and fluxes are difficult to measure and upscale due to their spatiotemporal heterogeneity. Therefore, it is important to combine the knowledge derived from experimental results with modeling. Here, we systematically compare data from a free air CO2 enrichment (FACE) experiment in a mature deciduous forest in Switzerland with realizations from an ecohydrological model (Tethys–Chloris). We test whether a mechanistic ecohydrological model is able to simulate physiological plant responses under ambient and elevated CO2 concentration. We overcome measurement limitations by quantifying differences in response to ambient and elevated CO2 over ten years. The reliability of model realizations is demonstrated by comparing simulations with field observations of stomatal conductance, sap flow, leaf and fruit litter, and stem growth. The model successfully captures the observed CO2-induced difference in stomatal conductance and transpiration and its sensitivity to atmospheric demand, as well as qualitative changes in soil moisture. The simulated differences between CO2 scenarios generally fall within the uncertainty of experimental observations, both for the carbon and water balance. Simulated total evapotranspiration is 2.8% (18mmyr−1) lower and soil moisture 1.2% higher in the CO2-enriched scenario. Latent and sensible heat are modified by ca. 1Wm−2. Net primary production is simulated to increase by 19.8% and allocation to stem growth is 53gCyr−1m−2 higher in the elevated CO2 scenario, which represents the limit of the detection threshold of the experiment. Results show that while ecohydrological models can be used to reliably simulate multi-year energy, water, and carbon fluxes at the stand level, testing carbon allocation remains critical with current accuracy of field measurements. Uncertainties due to the simplified carbon allocation scheme are shown to be more significant for carbon than for energy and water fluxes. Generally, we conclude that for this type of forest, differences in annual energy and water fluxes induced by elevated CO2 are likely to be less than 10%.