Four different extraction methods, soxhlet, soxtherm, sonication and accelerated solvent extraction (ASE), were used to isolate the 16 priority pollutant PAHs from a certified reference soil (LGC 6140) and from a contaminated soil (BG CLR 17). Based on SIM-GC-MS results, all methods were found to give accurate and highly reproducible concentration data. There was, however, significant between-method and sometimes within-method variability in the stable carbon isotope signatures obtained for individual PAHs from the contaminated soil (BG CLR 17) using GC-C-IRMS. When two clean-up procedures, silica/dichloromethane and alumina/hexane/toluene, were used to remove co-extracted material, however, it was found that ASE gave the more consistent and reproducible stable carbon isotope data. These findings are likely to be of importance for the characterisation of natural and anthropogenic organic matter and, in particular, in source identification and apportionment studies.