For 20 years, there has been 42,000 ha estate of clonal Eucalyptus plantations around Pointe-Noire in Congo on sandy soils that have very low reserves of available nutrients. These plantations have been based on a natural hybrid (E. PF1). This hybrid is being replaced by E. urophylla x E. grandis (UG), a more productive hybrid developed by the breeding program of UR2PI. A study of biogeochemical cycles showed that nutrient removal by harvesting is the main nutrient output in the E. PFI ecosystem. It is therefore important to quantify the nutrient content (NC) in both hybrids to compare corresponding nutrient removal values. The work dealt with four UG clones and the most planted clone of E. PFI. Twelve trees per clone were sampled at the logging age (8 years) in a clonal test for UG clones and in a nearby stand for E. PEI. Tables were established to predict, from girth at breast height (C1.30 m), the biomass and nutrient content of stemwood, bark, dead and living branches, leaves, and were applied to the inventory of the different stands to evaluate corresponding biomass, NC and nutrient use efficiency (NUE) on a per-hectare basis. Total biomass differed between the two hybrids and among UG clones: 109 t ha-1 for E. PFI and 108-155 t ha-1 for UG clones. In E. PF1 trees, total NC was globally lower for N, K, and Mg, but greater for P and Ca. In stemwood, nitrogen content was similar for both hybrids. By contrast, in UG clones, NC was much lower for P (-72%) and Ca (-40% to -55%). The same trends were observed for NUE: equivalent for both hybrids for N, but higher in UG clones for P (+72%) and Ca (+43% to +59%). A marked variability among clones was observed for K and Mg. UG clones allocated proportionally more nutrients in leaves than E. PF1. These results show that clones should not be selected only on growth traits but also on NUE and on the concentration of nutrients in tree components removed by harvesting. It will be then possible to limit the cost of fertilisin