Isoprene is a highly reactive volatile organic compound emitted by vegetation, known to be a precursor of secondary organic aerosols and to enhance tropospheric ozone formation under polluted conditions. Isoprene emissions respond strongly to changes in meteorological parameters such as temperature and solar radiation. In addition, the increasing CO2 concentration has a dual effect, as it causes both a direct emission inhibition as well as an increase in biomass through fertilization. In this study we used the MEGAN (Model of Emissions of Gases and Aerosols from Nature) emission model coupled with the MOHYCAN (Model of HYdrocarbon emissions by the CANopy) canopy model to calculate the isoprene fluxes emitted by vegetation in the recent past (1979–2014) and in the future (2070–2099) over Europe at a resolution of 0.1° × 0.1°. As a result of the changing climate, modeled isoprene fluxes increased by 1.1%yr−1 on average in Europe over 1979–2014, with the strongest trends found over eastern Europe and European Russia, whereas accounting for the CO2inhibition effect led to reduced emission trends (0.76%yr−1). Comparisons with field campaign measurements at seven European sites suggest that the MEGAN–MOHYCAN model provides a reliable representation of the temporal variability of the isoprene fluxes over timescales between 1h and several months. For the 1979–2014 period the model was driven by the ECMWF ERA-Interim reanalysis fields, whereas for the comparison of current with projected future emissions, we used meteorology simulated with the ALARO regional climate model. Depending on the representative concentration pathway (RCP) scenarios for greenhouse gas concentration trajectories driving the climate projections, isoprene emissions were found to increase by +7% (RCP2.6), +33% (RCP4.5), and +83% (RCP8.5), compared to the control simulation, and even stronger increases were found when considering the potential impact of CO2 fertilization: +15% (RCP2.6), +52% (RCP4.5), and +141% (RCP8.5). However, the inhibitory CO2 effect goes a long way towards canceling these increases. Based on two distinct parameterizations, representing strong or moderate inhibition, the projected emissions accounting for all effects were estimated to be 0–17% (strong inhibition) and 11–65% (moderate inhibition) higher than in the control simulation. The difference obtained using the two CO2parameterizations underscores the large uncertainty associated to this effect.