a b s t r a c tPoplar plantations have widely spread around the world due to its high productivity and adaptability. Clear-cutting is the primary harvesting method for poplar plantation management in southern China. However, the effect of harvesting on ecosystem carbon fluxes limits our ability to estimate its carbon sequestration. A consecutive, three-year observation on ecosystem CO 2 and CH 4 flux (F CO2 and F CH4 ) of a Populus dettoides plantation on the floodplain of Yangtze River was made prior and post to the clearcutting using an eddy-covariance system. We found that clear-cutting turned the ecosystem from a strong carbon sink to a mediate carbon source only in several months, July to next January, after the harvest. The ecosystem turned to a net carbon sink at the beginning of the first growing season following clear-cutting due to the large productivity of understory vegetation in this region. The annual carbon budget was À424.3 ± 52.5 g-C m À2 (95% confidence interval) in the harvesting year, with À53.6 ± 22.8 g-C m À2 the first year and À290.7 ± 34.2 g-C m À2 the second year after clear-cutting. Clear-cutting turned the ecosystem from a net CH 4 sink to a net CH 4 source after the third month, but during the three years the CH 4 emission only balanced out a very small portion (0.3%) of F CO2 . In noninundation periods, F CH4 varied from À0.01 to 0.24 mmol m À2 d À1 , with a mean (±SD) of 0.11 ± 0.08 mmol m À2 d À1 , while it ranged from 0.33 to 4.39 mmol m À2 d À1 during inundation, with a mean (±SD) of 2.17 ± 1.16 mmol m À2 d À1 . Daily and weekly F CH4 during non-inundation period were highly correlated with ground water table, soil moisture, and friction velocity, while F CH4 during inundation depended on inundation depth.