We compared radiation-use efficiency of growth (ɛ), defined as rate of biomass accumulation per unit of absorbed photosynthetically active radiation, of forest plots exposed to ambient (∼360 μl l−1) or elevated (∼560 μl l−1) atmospheric CO2 concentration ([CO2]). Large plots (30-m diameter) in a loblolly pine (Pinus taeda L.) plantation, which contained several hardwood species in the understory, were fumigated with a free-air CO2 enrichment system. Biomass accumulation of the dominant loblolly pines was calculated from monthly measurements of tree growth and site-specific allometric equations. Depending on the species, leaf area index (L*) was estimated by three methods: optical, allometric and litterfall. Based on the relationship between tree height and diameter during the first 3 years of exposure, we conclude that elevated [CO2] did not alter the pattern of aboveground biomass allocation in loblolly pine. There was considerable variation in L* estimates by the different methods; total L* was 18–42% lower when estimated by the optical method compared with estimates from allometric calculations, and this discrepancy was reduced when optical measurements were corrected for the non-random distribution of loblolly pine foliage. The allometric + litterfall approach revealed a seasonal maximum total L* of 6.2–7.1 with about 1/3 of the total from hardwood foliage. Elevated [CO2] had only a slight effect on L* in the first 3 years of this study. Mean ɛ (± SD), calculated for loblolly pine only, was 0.49 ± 0.05 and 0.62 ± 0.04 g MJ−1 for trees in the ambient and elevated [CO2] plots, respectively. The 27% increase in ɛ in response to CO2 enrichment was caused primarily by the stimulation of biomass increment, as there was only a small effect of elevated [CO2] on L* during the initial years of fumigation. Long-term increases in atmospheric [CO2] can increase ɛ in closed-canopy forests but the absolute magnitude and duration of this increase remain uncertain.