More reliable estimates of carbon (C) stock within forest ecosystems and C emission induced by deforestation are urgently needed to mitigate the effects of emissions on climate change. A process-based terrestrial biogeochemical model (VISIT) was applied to tropical primary forests of two types (a seasonal dry forest in Thailand and a rainforest in Malaysia) and one agro-forest (an oil palm plantation in Malaysia) to estimate the C budget of tropical ecosystems, including the impacts of land-use conversion, in Southeast Asia. Observations and VISIT model simulations indicated that the primary forests had high photosynthetic uptake: gross primary production was estimated at 31.5–35.5 t C ha⁻¹ yr⁻¹. In the VISIT model simulation, the rainforest had a higher total C stock (plant biomass and soil organic matter, 301.5 t C ha⁻¹) than that in the seasonal dry forest (266.5 t C ha⁻¹) in 2008. The VISIT model appropriately captured the impacts of disturbances such as deforestation and land-use conversions on the C budget. Results of sensitivity analysis implied that the ratio of remaining residual debris was a key parameter determining the soil C budget after deforestation events. The C stock of the oil palm plantation was about 46% of the rainforest's C at 30 yr following initiation of the plantation, when the ratio of remaining residual debris was assumed to be about 33%. These results show that adequate forest management is important for reducing C emission from soil and C budget of each ecosystem must be evaluated over a long term using both the model simulations and observations.