The 13C natural abundance of CO2 respired by plants has been used in the laboratory to examine the discrimination processes that occur during respiration. Currently, field measurements are being expanded to interpret the respiration δ13C signature measured at ecosystem and global levels. In this context, forests are particularly important to consider as they represent 80% of the continental biomass. The objective of this investigation was to compare four methods of sampling the CO2 respired by trunks for the determination of its carbon isotope composition: three in situ methods using chambers placed on the trunk, and one destructive method using cores of woody tissues. The in situ methods were based either on a Keeling plot approach applied at the tissue level or on an initial flush of the chamber with nitrogen or with CO2-free air. In parallel, we investigated the possibility of an apparent discrimination during tissue respiration by comparing the δ13C signature of the respired CO2 and that of the organic matter. The study was performed on six tree species widely distributed in temperate and mediterranean areas. The four methods were not significantly different when overall means were considered. However, considering the individual data, the Keeling plot approach and the nitrogen flush methods gave fairly homogeneous results, whereas the CO2-free air method produced more variable results. The core method was not correlated with any of the chamber methods. Regardless of the methodology, the respired CO2 generally was enriched in 13C relative to the total organic matter. This apparent enrichment during respiration was variable, reaching as much as 3–5‰. This study showed that, on the whole, the different sampling techniques gave similar results, but one should be aware of the variability associated with each method. Copyright © 2005 John Wiley & Sons, Ltd.