Abstract This article contributes to the literature by investigating the dynamic relationship between carbon dioxide (CO2) emissions, output (GDP), energy consumption, and trade using the bounds testing approach to cointegration and the ARDL methodology for Tunisia over the period 1971–2008. The empirical results reveal the existence of two causal long-run relationships between the variables. In the short-run, there are three unidirectional Granger causality relationships, which run from GDP, squared GDP and energy consumption to CO2 emissions. To check the stability in the parameter of the selected model, CUSUM and CUSUMSQ were used. The results also provide important policy implications.