We used a continuous 9 year (2003–2011) eddy flux time series with 30 min resolution to examine water use efficiency in a tropical rainforest and determine its environmental controls. The multiyear mean water use efficiency (W_ue) of this rainforest was 3.16 ± 0.33 gC per kg H₂O, which is close to that of boreal forests, but higher than subtropical forests, and lower than temperate forests. The water vapor deficit (V_PD) had a strong impact on instantaneous W_ue, in the manner predicted by stomatal optimization theory. At the seasonal scale, temperature was the dominant controller of W_ue. The negative correlation between temperature and W_ue was probably caused by high continuous photosynthesis during low‐temperature periods. The V_PD did not correlate with W_ue at the interannual scale. No interannual trend was detected in W_ue or inherent water use efficiency (W_ei), either annually or seasonally. The fact that no increasing trend of W_ei was found in the studied tropical rainforest, along with other evidence of CO₂ stimulation in tropical rainforests, requires special attention and data validation. There was no significant difference between W_ue during a drought and the 9 year mean values in the forest we studied, but we found that dry season transpiration (T_r) was consistently lower during the drought compared to the mean values. Finally, whether W_ue increases or decreases during a drought is determined by the drought sensitivity of gross primary production (G_PP).

The response of tropical rainforest to environmental perturbation is of critical importance given their potential to mitigate climate change. Nevertheless, it was not well addressed to date. Therefore, related hypotheses, i.e., CO₂ fertilization–related accelerating growth hypothesis (AGH) and remote sensing–based drought resilience hypothesis (DRH), were necessarily to be testified. Here these hypotheses were tested through 10 years of annual inventory records and half‐hourly eddy flux measurements from a tropical rainforest. We show that the studied forest is highly sensitive to water variability, with low canopy photosynthesis, slow stand growth, and high mortality rate in dry years, especially in the severe drought. Ecosystem respiration was not correlated with net water balance within years, but significant correlations were found between these parameters with a time lag of 10–15 months. A boom of photosynthesis in 1 year post drought was most probably a result of nutrient pulse–related drought. In general, neither AGH nor DRH was supported by the study. Given predictions that tropical areas will experience increasingly dry conditions in the future, much attention should be paid to the potential fate of carbon sink in tropical rainforests.