In addition to its important effect on soil quality and crop productivity, soil organic carbon has also been identified as a possible C sink for sequestering atmospheric carbon dioxide. The Rothamsted carbon model, RothC, has been widely tested in many countries. The source of carbon input into soils consists of below-ground and above-ground plant biomass. The model requires users to estimate carbon inputs, rather than being directly simulated by the model. Measured above-ground dry matter and crop yields are often used to estimate the amount of carbon input. Theoretically, if above-ground stubble is neither removed from the field nor burnt at the site, the stubble retention factor is set as 1.0, otherwise, it is 0.0 or a value between 0.0 and 1.0, depending on the proportion left in the field. In this study, the RothC model is used to simulate SOC in a long-term field experiment (1979'2004) near Wagga Wagga, NSW, which involved different crop rotation, tillage and stubble management practices. The simulation showed that modelled SOC adequately matched observed SOC in the stubble burnt treatments, but failed to simulate the stubble retained treatments. The root mean squared errors (RMSE) were 2.86 and 2.18 t C ha 1 for no-tillage (NT) and cultivated treatments (CC) when stubble was burnt, and 9.93 (NT) and 11.51 t C ha 1 (CC) when stubble was retained, respectively. Using an inverse simulation technique, results showed that when the stubble retention factor was 26%, the simulated SOC can match well the observed SOC, yielding RMSE of 2.33 t C ha 1 or lower. The results suggested that a large fraction of the surface-retained stubble was lost before entering soil to be sequestrated as SOC.