Coordinate rotation is an essential step in eddy covariance (EC) flux measurements particularly on complex terrains. Systematically assessing impacts of coordinate rotation methods can improve the accuracy in EC flux estimations. Taking the Maoershan flux tower in a temperate deci-duous forest on valley slopes as a case, we comprehensively examined the effects of various tilt-correction methods on sensible heat flux (H), latent heat flux (LE), CO2 fluxes (Fc), energy ba-lance, friction velocity (u*), and vertical velocity (w) using the standardized major axis method. Compared with the unrotated fluxes, the double rotation (DR), planar fit (PF), vertical velocity unbiased planar fit (NBPF), monthly planar fit (MPF), triple rotation (TR) and monthly sector-wist planar fit (MSWPF) changed the H by +1.5%, +3.8%, +3.1%, +3.3%, -1.3% and -7.1%, respectively. The NBPF and DR increased LE by 1.0% and 0.4%, respectively. The PF also increased LE by 0.671 W·m-2, and the MPF decreased LE by 0.4%, but increased the intercept by 0.747 W·m-2. In contrast, the MSWPF and TR decreased the LE by 4.5% and 15.1%, respectively. The coordinate rotation reduced the Fc by 6.0% (DR)-12.1% (TR). Among the studied methods, the PF increased the energy balance closure by about 2%, whereas the TR decreased it by about 6%. In the light of u* and w, the TR and NBPF were inappropriate for anemometer tilt correction. Considering the precision and practical application of EC measurements, we recommend the PF as the first choice for the Maoershan flux tower site, followed by the DR. These findings are useful for tilt correction of anemometer in EC flux observations over complex terrains.