Ecosystem water use efficiency (WUE) reflects the intimately coupled relationship between the carbon and water cycles in terrestrial ecosystems. However, the inter-annual variation and its drivers of WUE are poorly understood at regional/global scale, due to either limited data availability or uncertainties in current data streams. In this study, we used process-based models simulated gross primary productivity (GPP) and evapotranspiration (ET) data to estimate the ecosystem WUE (eWUE, GPP/ET) in China for 1979-2012. The eWUE estimates were validated against eddy covariance-based values from 35 flux towers. The inter-annual variation of the eWUE was quantified and its responses to annual precipitation (AP), annual mean temperature (AMT), and annual mean leaf area index (AMLAI) were analyzed. The key findings were as follows. (i) The mean annual eWUE over China was 1.48 +/- 1.04 g C kg(-1) H2O and had a slightly increasing but not significant trend (7.32 x 10(-4)g C kg(-1) H2O yr(-1), p < 0.05) from 1979 to 2012. (ii) The spatial distribution of the eWUE trend showed large spatial variability. similar to 21.4% and similar to 19.0% of vegetated land in China had significant increasing and decreasing trends (Mann-Kendall test, p < 0.1), respectively. The increasing eWUE was mainly found in the northeast, southwest, and central areas of China, while the decreasing eWUE was mostly distributed in west China. (iii) The inter annual variation of the spatially averaged annual eWUE was negatively correlated with that of AP and AMT, and positively correlated with that of AMLAI. In similar to 41.4%, similar to 9.9%, and similar to 3.1% of vegetated land in China the inter annual variation of eWUE was dominated by the inter-annual variations of AP, AMT, and AMLAI, respectively. In most land of north China and west China the inter-annual variation of eWUE was dominated by the inter-annual variation of AP, while in central, east and south China all the AP, AMT, AMLAI, and other drivers played important roles.