Sulfur (S) isotope ratios of thoroughly dried organic samples were measured by direct thermal decomposition in an elemental analyzer coupled to an isotope ratio mass spectrometer in continuous flow mode (EA-CF-IRMS). For organic samples of up to 13 mg weight and with total S contents of more than 10 μg, the reproducibility of the δ34Sorganic values was ±0.4‰ or better. However, the δ34S values of organic samples measured directly by online EA-CF-IRMS analysis were between 0.3 and 2.9‰ higher than those determined on BaSO4 precipitates produced by Parr Bomb™ oxidation from the same sample material. Our results suggest that structural oxygen in organic samples influences the oxygen isotope ratios of the SO2 produced from organic samples. Consequently, SO2 generated from organic samples appears to have different 18O/16O ratios than SO2 generated from BaSO4 precipitates and inorganic reference materials, resulting in a deviation from the true δ34S values because of 32S16O18O contributions to mass 66. It was shown that both the amount of structural oxygen in the organic sample, and the difference of the oxygen isotope ratios between organic samples and tank O2, influenced the magnitude of the observed deviation from the true δ34S value after direct EA-CF-IRMS analysis of organic samples. Suggestions are made to correct the difference between measured δ34Sorganic and true δ34S values in order to obtain not only reproducible, but also accurate S isotope ratios for organic materials by EA-CF-IRMS. Copyright © 2005 John Wiley & Sons, Ltd.