Experimentation with dynamics of soil carbon pools as affected by elevated CO2 can better define the ability of terrestrial ecosystems to sequester global carbon. In the present study, 6 N HCl hydrolysis and stable-carbon isotopic analysis (δ13C) were used to investigate labile and recalcitrant soil carbon pools and the translocation among these pools of sorghum residues isotopically labeled in the 1998-1999 Arizona Maricopa free air CO2 enrichment (FACE) experiment, in which elevated CO2 (FACE: 560 μmol mol-1) and ambient CO2 (Control: 360 μmol mol-1) interact with water-adequate (wet) and water-deficient (dry) treatments. We found that on average 53% of the final soil organic carbon (SOC) in the FACE plot was in the recalcitrant carbon pool and 47% in the labile pool, whereas in the Control plot 46% and 54% of carbon were in recalcitrant and labile pools, respectively, indicating that elevated CO2 transferred more SOC into the slow-decay carbon pool. Also, isotopic mixing models revealed that increased new sorghum residue input to the recalcitrant pool mainly accounts for this change, especially for the upper soil horizon (0-30 cm) where new carbon in recalcitrant soil pools of FACE wet and dry treatments was 1.7 and 2.8 times as large as that in respective Control recalcitrant pools. Similarly, old C in the recalcitrant pool under elevated CO2 was higher than that under ambient CO2, indicating that elevated CO2 reduces the decay of the old C in recalcitrant pool. Mean residence time (MRT) of bulk soil carbon at the depth of 0-30 cm was significantly longer in FACE plot than Control plot by the averages of 12 and 13 yr under the dry and wet conditions, respectively. The MRT was positively correlated to the ratio of carbon content in the recalcitrant pool to total SOC and negatively correlated to the ratio of carbon content in the labile pool to total SOC. Influence of water alone on the bulk SOC or the labile and recalcitrant pools was not significant. However, water stress interacting with CO2 enhanced the shift of the carbon from labile pool to recalcitrant pool. Our results imply that terrestrial agroecosystems may play a critical role in sequestrating atmospheric CO2 and mitigating harmful CO2 under future atmospheric conditions.