The effect of doubling atmospheric CO2 concentration (C-a) on climate and vegetation is investigated using a combined climate-vegetation model. The vegetation model predicts the response of leaf area index, canopy transpiration (E(T)) and whole-plant carbon balance to changes in climate, soil moisture, and atmospheric CO2 forcing. This model has been embedded in the UK Meteorological Office Single Column Model (SCM), which provides the climate feedback to the vegetation. The vegetation model uses an optimisation approach to predict stomatal resistance, a biochemical model to predict photosynthesis and a simple carbon balance model to predict leaf area. Respiration is calculated as a function of leaf area and vegetation height. Clouds are assumed to be radiatively passive in the SCM to avoid unrealistic feedbacks. Simulations were performed with the fully interactive vegetation-climate model for an Amazon location with the present-day value of C-a (1 x CO2), and twice this value (2 x CO2). In addition, two other types of simulation were performed at both CO2 concentrations: one in which the vegetation component was forced only with 1 x CO2, and one using a fixed surface resistance. The latter case is equivalent to simulations using most current general circulation models. In all the simulations, increased atmospheric CO2 caused an increase in surface temperature owing to increased radiative forcing. With a fixed resistance, mean E(T) was increased by 5.6% and sensible heat flux was reduced by 3.8%. The fully interactive model had significant effects on the response of both climate and productivity to C-a. Increased C-a caused stomatal closure, which resulted in a reduction in mean E(T) Of 25%. The effect of C-a on E(T) was amplified by the positive feedback resulting from the effect of increased air humidity deficit on stomatal resistance.