Compared to conventional abatement measures, Reducing Emissions from Deforestation and Forest Degradation (REDD) offers attractive cost savings while tackling the climate change problem. However, there exist challenges associated with the selection of the optimal level of REDD-based abatement given the risks and non-uniform costs of their implementation across countries. This paper develops an integrated assessment model of carbon mitigation, incorporating the REDD option. Using a dynamic optimization framework, it derives the optimal timing and level of REDD participation for key countries with REDD potential based on their opportunity costs and risks. Specifically, Brazil, Indonesia, the Democratic Republic of Congo, Cameroon and Papua New Guinea, are chosen for inclusion under the REDD-based abatement option. Together, these five countries account for roughly 20 percent of global forest area and 40 percent of current global deforestation. The relevance and contribution of REDD-based abatement is explored under the possibility of non-linear damages resulting from increasing concentrations of greenhouse gases. Results indicate that the REDD programme is an attractive option to consider despite the associated risks of impermanence. Including the REDD option, in fact, also increases conventional abatement efforts because low costs of REDD reduce the overall abatement costs, thereby making it optimal to abate more. Further, use of REDD option helps stabilise the atmospheric carbon stock in the long term. Without REDD, atmospheric carbon concentrations would be higher by 800 billion tonnes in the next 300 years. Whereas, optimal implementation of REDD in just five countries would help avoid the release of about 80 billion tonnes of carbon in the next 50 years.