This study investigates the influence of interannual vegetation variability. Two sets of offline and online simulations were performed using the Community Earth System Model. The interannual Global LAnd Surface Satellite (GLASS) leaf area index (LAI) dataset from 1985 to 2000 and its associated climatological LAI were used to replace the default climatological LAI data in version 4 of the Community Land Model (CLM4). The results showed that on a global scale, canopy transpiration and evaporation, as well as total evapotranspiration in offline simulations were significantly positively correlated with LAI, whereas ground evaporation and ground temperature showed significant negative correlation with LAI. However, the correlations in online simulations were reduced markedly because of interactive feedbacks between albedo, changed climatic factors and atmospheric variability. In the offline simulations, the fluctuations of differences in interannual variability of evapotranspiration and ground temperature focused on vegetation growing regions and the magnitudes were smaller. Those in online simulations spread over more regions and the magnitudes were larger. These results highlight the influence of interannual vegetation variability, particularly in online simulations, an effect that deserves consideration and attention when investigating the uncertainty of climate change.