How forests will respond to rising [CO2] in the long term is uncertain, most studies having involved juvenile trees in chambers prior to canopy closure. Poplar free-air CO2 enrichment (Viterbo, Italy) is one of the first experiments to grow a forest from planting through canopy closure to coppice, entirely under open-air conditions using free-air CO2 enrichment technology. Three Populus species: P. alba, P. nigra and P. x euramericana, were grown in three blocks, each containing one control and one treatment plot in which CO2 was elevated to the expected 2050 concentration of 550 ppm. The objective of this study was to estimate gross primary production (GPP) from recorded leaf photosynthetic properties, leaf area index (LAI) and meteorological conditions over the complete 3-year rotation cycle. From the meteorological conditions recorded at 30 min intervals and biweekly measurements of LAI, the microclimate of leaves within the plots was estimated with a radiation transfer and energy balance model. This information was in turn used as input into a canopy microclimate model to determine light and temperature of different leaf classes at 30 min intervals which in turn was used with the steady-state biochemical model of leaf photosynthesis to compute CO2 uptake by the different leaf classes. The parameters of these models were derived from measurements made at regular intervals throughout the coppice cycle. The photosynthetic rates for different leaf classes were summed to obtain canopy photosynthesis, i.e. GPP. The model was run for each species in each plot, so that differences in GPP between species and treatments could be tested statistically. Significant stimulation of GPP driven by elevated [CO2] occurred in all 3 years, and was greatest in the first year (223–251%), but markedly lower in the second (19–24%) and third years (5–19%). Increase in GPP in elevated relative to control plots was highest for P. nigra in 1999 and for P. x euramericana in 2000 and 2001, although in 1999 P. alba had a higher GPP than P. x euramericana. Our analysis attributed the decline in stimulation to canopy closure and not photosynthetic acclimation. Over the 3-year rotation cycle from planting to harvest, the cumulative GPP was 4500, 4960 and 4010 g C m−2 for P. alba, P. nigra and P. x euramericana, respectively, in current [CO2] and 5260, 5800 and 5000 g C m−2 in the elevated [CO2] treatments. The relative changes were consistent with independent measurements of net primary production, determined independently from biomass increments and turnover.