This study investigated the relationship between delta13C of ecosystem components, soluble plant carbohydrates and the isotopic signature of ecosystem respired CO2 (delta13CR) during seasonal changes in soil and atmospheric moisture in a beech (Fagus sylvatica L.) forest in the central Apennine mountains, Italy. Decrease in soil moisture and increase in air vapour pressure deficit during summer correlated with substantial increase in delta13C of leaf and phloem sap soluble sugars. Increases in delta13C of ecosystem respired CO2 were linearly related to increases in phloem sugar delta13C (r2=0.99, P<or=0.001) and leaf sugar delta13C (r2=0.981, P<or=0.01), indicating that a major proportion of ecosystem respired CO2 was derived from recent assimilates. The slopes of the best-fit lines differed significantly (P<or=0.05), however, and were about 0.86 (SE=0.04) for phloem sugars and about 1.63 (SE=0.16) for leaf sugars. Hence, changes in isotopic signature in phloem sugars were transferred to ecosystem respiration in the beech forest, while leaf sugars, with relatively small seasonal changes in delta13C, must have a slower turnover rate or a significant storage component. No significant variation in delta13C was observed in bulk dry matter of various plant and ecosystem components (including leaves, bark, wood, litter and soil organics). The apparent coupling between the delta13C of soluble sugars and ecosystem respiration was associated with large apparent isotopic disequilibria. Values of delta13CR were consistently more depleted by about 4 per thousand relative to phloem sugars, and by about 2 per thousand compared to leaf sugars. Since no combination of the measured pools could produce the observed delta13CR signal over the entire season, a significant isotopic discrimination against 13C might be associated with short-term ecosystem respiration. However, these differences might also be explained by substantial contributions of other not measured carbon pools (e.g., lipids) to ecosystem respiration or contributions linked to differences in footprint area between Keeling plots and carbohydrate sampling. Linking the seasonal and inter-annual variations in carbon isotope composition of carbohydrates and respiratory CO2 should be applicable in carbon cycle models and help the understanding of inter-annual variation in biospheric sink strength.