Afforestation in arid land is a promising method for carbon fixation, but the effective utili-zation of water is highly important and required. Thus, the evaluation of the amount of water per unit carbon fixed with the tree growth is required to minimize the amount of water supplied to the plants. In this research, a tree is regarded as a carbon fixation reactor with inflows of water andnutrients from roots, and CO_2 as the carbon source from leaves with outflow of water vapor from leaves and accumulation in the tree itself. In the process of photosynthesis and respiration nutri-tional elements are dissolved in water flow in trees. They do not flow out by these reactions, but are accumulated in trees. Thus, we have treated the behaviour of nutrients as a marker to evaluate the water/carbon ratio. Assuming that nutrient concentration is constant in sap, and the differences in the ratios of nutrient to carbon in living trees and dead (i.e. litter fall, etc.) are neglected, the ratio of the usedwater to fixed carbon is given as the ratio of nutrient to carbon in the tree body divided by the ratioof nutrient to water in sap. However, some nutrients are translocated and concentrated within the tree and some may be discarded through litter fall. Thus it is important to examine which nutrient element is the most suitable as the tracer. In this paper, the results of the above method applied to Eucalyptus camaldulensis in semi-arid land of Western Australia are shown. The value of water requirement per unit carbon fixation determined from potassium balance is between 421 kg-H_2O/kg-C for mature trees and 285kg-H_2O/kg-C for young trees, while the values from calcium balance are much larger than these. The cause of the discrepancy between these values is discussed based on the measured element concentrations in sap and trees and the plant physiology. Finally, the actual average value throughthe life of a tree is suggested to fall between the two values. Afforestation in arid land is a promising method for carbon fixation, but the effective utili-zation of water is highly important and required. Thus, the evaluation of the amount of water per unit carbon fixed with the tree growth is required to minimize the amount of In this research, a tree is considered as a carbon fixation reactor with inflows of water and nutrients from roots, and CO_2 as the carbon source from leaves with outflow of water vapor from leaves and accumulation in the tree itself. In the process of photosynthesis and respiration nutri-tional elements are dissolved in water flow in trees. Thus, we have treated the behavior of nutrients as a marker to evaluate the water / carbon ratio. Assuming that nutrient concentration is constant in sap, and the differences in the ratios of nutrient to carbon in living trees and dead (ie litter fall, etc.) are neglected, the ratio of t he usedwater to fixed carbon is given as the ratio of nutrient to carbon in the tree body divided by the ratio of nutrient to water in sap. Yet it some is this important to examine which nutrient element is the most suitable as the tracer. In this paper, the results of the above method applied to Eucalyptus camaldulensis in semi-arid land of Western Australia are shown. from potassium balance is between 421 kg-H_2O / kg-C for mature trees and 285 kg-H_2O / kg-C for young trees, while the values ​​from calcium balance are much more than these. The cause of the discrepancy between these values ​​is discussed based on the measured element concentrations in sap and trees and the plant physiology. Finally, the actual average value through the life of a tree is suggested to fall between the two values.