Abstract The performance of a gas chromatography‐combustion‐isotope ratio mass spectrometry system (GC‐C‐IRMS) with respect to the dependence of δ 13 C values on the amount of sample is presented. Particular attention is paid to the localization of the amount‐dependent isotopic fractionation within the system. Injection experiments with varying amounts of gases (CO 2 , n‐hexane, and toluene) revealed that neither the detector unit nor the combustion reactor, but rather the conditions in the split/splitless injector, contributed to this effect. Although optimization of injector parameters was performed and a reduction of this adverse effect from 3 to 1‰ was achieved, it was not possible to eliminate isotopic fractionation completely. Consequently, additional injector parameters have to be considered and adjusted to achieve injection conditions free of fractionation. For routine analysis of the compound‐specific δ 13 C analysis of different biomarkers in many environmental samples, perfect optimization may not always be reached. Therefore, in order to prevent systematic errors in the measured δ 13 C values due to different sample concentrations, it is suggested that correction for the remaining unknown amount‐dependent fractionation can be made by means of co‐analyzing standards of varying analyte concentrations and known δ 13 C values. Residual overall amount‐dependent isotope‐fractionation can thus be corrected mathematically. Copyright © 2003 John Wiley & Sons, Ltd.