We tested the ability of the Rothamsted Carbon Model (RothC-26.3) to predict current soil organic carbon (SOC) levels in three long-term (23-25 years old) experimental sites in Nepal (Bhairahawa, Parwanipur, and Tarhara)  under rice-rice-wheat (RRW) or rice-wheat (RW) systems for different nutrient management scenarios. The model was also evaluated to simulate the regional pattern of SOC for  soils with different soil textures  under  RW system in the Indo-Gangetic Plains  of South Asia. The model outputs for changes in SOC were compared with measured data in the long-term fertility experiments and RW farmers' fields in surface soil layer. The inert organic matter pool was set at 75% of the value obtained from a regression equation for the minimum observed SOC as a function of soil texture. In general, amounts of soil organic carbon predicted by the model agreed fairly well with the amounts measured for all treatments (unfertilized, NPK, farm yard manure (FYM), FYM+NPK, and chopped wheat straw + NPK) at Bhairahawa and Parwanipur experiments but overestimated  SOC at Tarhara, especially for the FYM treatment. The modeled and measured results indicated that unfertilized plots led to a continuous decline in SOC during the study period and chemical fertilizer alone was inadequate to maintain the SOC levels in the plow layer. The results also showed that the model adequately represented the effect of soil texture on SOC levels. RothC does not seem to be limited by a lack of rate constant adjustment for anaerobic conditions but may well need a tillage parameter.  Our results confirm that RothC model  performs well in subtropical region  under rice (anaerobic phase)-wheat (aerobic phase) systems.