Adaptive strategies to improve tree water-use efficiency (WUE) are required to meet the global demand for wood in a future drier climate. A large-scale throughfall exclusion experiment was set up in Brazil to study the interaction between water status and potassium (K) or sodium (Na) availability on the ecophysiology of Eucalyptus grandis trees. This experiment focused primarily on the changes in aboveground net primary production, stand water use, phloem sap and leaf d13C, net CO2 assimilation and stomatal conductance. The correlations between these response variables were determined to gain insight into the factors controlling water-use efficiency in tropical eucalypt plantations. The intrinsic WUE in individual leaves (the ratio of net CO2 assimilation to stomatal conductance) was estimated at a very short time scale from the leaf gas exchange. Sap flow measurements were carried out to assess the WUE for stemwood production (the ratio of wood biomass increment to stand water use). Averaged over the two water supply regimes, the stemwood biomass 3 years after planting was 173% higher in trees fertilized with K and 79% higher in trees fertilized with Na than in trees with no K and Na addition. Excluding 37% of the throughfall reduced stemwood production only for trees fertilized with K. Total canopy transpiration between 1 and 3 years after planting increased from about 750 to 1300 mm y-1 in response to K fertilization with a low influence of the water supply regime. K fertilization increased WUE for stemwood production by approx. 60% with or without throughfall exclusion. There was a strong positive correlation between phloem sap d13C and short-term leaf-level intrinsic WUE. Whatever the water and nutrient supply regime, the gas exchange WUE estimates were not correlated with WUE for stemwood production. Phloem sap d13C and leaf d13C were therefore not valuable proxies of WUE for stemwood production. The allocation pattern in response to nutrient and water supply