Detrending correction and sonic anemometer tilt correction were made to modify the raw time series measured from eddy covariance system in broad-leaved Korean pine forest of Changbai Mountains during the growing season of 2003, and the impact of different correction methods on CO2 flux was analyzed quantificationally. The results showed that the forest CO2 flux during growing season was overestimated when calculated from raw time series. The ratio of correction to origin flux (Fc(raw)) was 1.6% and 1.8% for linear and nonlinear detrend, respectively, which suggested that there was little difference between these two detrending methods. It was 3.7% and 4.7% for the planar fit coordinate transforming (PF) correction and the streamline coordinate system transforming (ST) correction, respectively, suggesting that there was a clear difference between these two sonic anemometer tilt correction methods. When detrended time series used, it was 5.5% and 4.6% for ST correction and PF correction, respectively. It was recommended that raw time series should be corrected synthetically with linear detrend method and PF method.