The paper reports a re-examination of the environmental Kuznets curve (EKC) hypothesis for Bangladesh; considering economic growth and greenhouse gas (GHG) emissions (such as CO2, CH4, and N2O). The autoregressive distributive lag (ARDL) model and VAR-based innovative accounting approach (IAA) with the combination of Variance Decomposition and Impulse Response Function are used to check cointegration among the variables using data from 1976 to 2014. This approach is specifically employed for variance analysis in order to measure causal association and regressors’ shock to dependent variables. The EKC hypothesis is not supported because there is a positive association between economic growth and pollutants (CO2 and CH4 emissions) over time. However, the EKC is valid for N2O emissions. IAA causal analysis reveals a bilateral causative relationship between GDP and CO2 emissions, as well as a unilateral causal relationship between CH4 emissions and GDP. Trade liberalisation, urbanisation, and financial liberalisation do not necessarily improve or save the natural environment. Bangladesh needs to use renewable energies and cut GHG emissions by abandoning fossil fuels.