The performance of the Rothamsted Carbon Model (RothC) in simulating soil carbon (SOC) storage in cotton based cropping systems under different tillage management practices on an irrigated Vertisol in semi-arid, subtropics was evaluated using data from a long-term (1994-2012) cotton cropping systems experiment near Narrabri in north-western New South Wales, Australia. The experimental treatments were continuous cotton/conventional tillage (CC/CT), continuous cotton/minimum tillage (CC/MT), and cotton-wheat (Triticum aestivum L.) rotation/minimum tillage (CW/MT). Soil carbon (C) input was calculated by published functions that relate crop yield to soil C input. Measured values showed a loss in SOC of 34%, 24% and 31% of the initial SOC storages within 19 years (1994-2012) under CC/CT, CC/MT, and CW/MT, respectively. RothC satisfactorily simulated the dynamics of SOC in cotton based cropping systems under minimum tillage (CC/MT and CW/MT), whereas the model performance was poor under intensive conventional tillage (CC/CT). The model RothC overestimated SOC storage in cotton cropping under conventional intensive tillage management system. This over estimation could not be attributed to the overestimation of soil C inputs, or errors in initial quantification of SOC pools for model initialization, or the ratio of incoming decomposable plant materials to resistant plant materials. Among other different factors affecting SOC dynamics and its modelling under intensive tillage in tropics and sub-tropics, we conclude that factors for tillage and soil erosion might be needed when modelling SOC dynamics using RothC under intensive tillage management system in the tropics and the sub-tropics.