In Venezuela 30,000 km2 of land is covered by savannas, of which 410,000 ha have been planted with several species and hybrids of Eucalyptus for lumber and pulp production. Popular concern about possible diminutions in water availability of reservoirs near eucalypt plantations prompted our interest in measuring water use by these trees. Since these savannas are markedly seasonal, the response of species to seasonal drought is important. We aimed to compare the seasonal changes in single-leaf and whole-plant transpiration in a seasonally dry savanna with that of trees of E. urophylla in an experimental plantation. We also examined the seasonal changes in xylem water potential and stomatal response to air water-vapour saturation deficit (D). Transpiration in eucalypts and the dominant savanna species Trachypogon vestitus and Curatella americana was evaluated using measurements of leaf gas exchange in all three species, sap flux in eucalypts, microclimatic variables and allometric and photometric determinations of green area. In E. urophylla and T. vestitus, but not in C. americana, stomatal conductance (gs) proved sensitive to D. Integrated values of daily courses of transpiration rate were scaled to one ha in a preliminary approach to estimating ecosystem transpiration (Eha). The Eha of the savanna (the sum of Eha of T. vestitus and C. americana) was on average 2. 4 times that of eucalypts during the daytime; when nocturnal eucalypt transpiration was included, the value was 1. 9. The evapotranspiration calculated by the Penman-Montieth equation (ETc) of eucalypts was lower than the savanna all year round. The reference crop ET (ETo) varied little throughout the seasons, the highest value occurring in March. The ratio Eha/ETo for the savanna was on average near one during the dry season and almost two during the rainy season; the corresponding value for E. urophylla was 0. 6 for both seasons. The ratio Eha/ETc was on average 0. 8 for the species and the savanna. The cumulative Eha for the days of measurements was higher in the savanna than in the eucalypts during the daytime (39. 8 and 17. 3 mm, respectively), as was the cumulative ETc (37. 5 vs. 20. 3 mm). Measured and calculated cumulative ET in eucalypts, including nocturnal values, were 22. 0 and 28. 4 mm, respectively. At the leaf level, both eucalypts and trees of C. americana apparently may have accessed water from deep horizons, since their values of Ψ changed less seasonally than in T. vestitus. At the ecosystem level, the species that presented the largest changes in transpiration was T. vestitus, which markedly increased savanna transpiration during the rainy season. Our results suggest that, for the days of this study, and considering the environmental conditions of the ecosystems studied, the type of measurements and the scaling procedures, stands of E. urophylla transpire less water than the savanna.