The Thailand Development Policy focuses on the simultaneous growth of the economy, society, and environment. Long-term goals have been set to improve economic and social well-being. At the same time, these aim to reduce the emission of CO2 in the future, especially in the construction sector, which is deemed important in terms of national development and is a high generator of greenhouse gas. In order to achieve national sustainable development, policy formulation and planning is becoming necessary and requires a tool to undertake such a formulation. The tool is none other than the forecasting of CO2 emissions in long-term energy consumption to produce a complete and accurate formulation. This research aims to study and forecast energy-related carbon dioxide emissions in Thailand’s construction sector by applying a model incorporating the long- and short-term auto-regressive (AR), integrated (I), moving average (MA) with exogenous variables (Xi) and the error correction mechanism (LS-ARIMAXi-ECM) model. This model is established and attempts to fill the gaps left by the old models. In fact, the model is constructed based on factors that are causal and influential for changes in CO2 emissions. Both independent variables and dependent variables must be stationary at the same level. In addition, the LS-ARIMAXi-ECM model deploys a co-integration analysis and error correction mechanism (ECM) in its modeling. The study’s findings reveal that the LS-ARIMAXi ( 2 , 1 , 1 , X t − 1 ) -ECM model is a forecasting model with an appropriate time period (t − i), as justified by the Q-test statistic and is not a spurious model. Therefore, it is used to forecast CO2 emissions for the next 20 years (2019 to 2038). From the study, the results show that CO2 emissions in the construction sector will increase by 37.88% or 61.09 Mt CO2 Eq. in 2038. Also, the LS-ARIMAXi ( 2 , 1 , 1 , X t − 1 ) -ECM model has been evaluated regarding its performance, and it produces a mean absolute percentage error (MAPE) of 1.01% and root mean square error (RMSE) of 0.93% as compared to the old models. Overall, the results indicate that determining future national sustainable development policies requires an appropriate forecasting model, which is built upon causal and contextual factors according to relevant sectors, to serve as an important tool for future sustainable planning.