The issue of urbanization has gained much importance over the last few decades due to its significant influence on economic growth and environmental quality, especially in developing countries. The non-linearity puzzle has been a long-debated issue, and prior studies provide mixed evidence. This study addresses the issue of urbanization using the measure of urban agglomeration and investigates the non-linear relation between urbanization and CO2 emissions at the regional level. The South Asian region represents approximately one-fourth of the world population and its urbanization needs to be addressed properly. This paper uses the annual data over the period of 1974–2014 for four South Asian countries, namely, Pakistan, India, Bangladesh, and Nepal. The panel cointegration tests establish the long-run relation between urbanization, urban agglomeration, economic growth, trade openness, energy consumption, financial development, and CO2 emissions. The fully modified ordinary least squares (FMOLS) model further confirms the existence of Environmental Kuznets Curve (EKC) in South Asia. Moreover, urbanization has an inverted U-shaped relation with CO2 emissions, while urban agglomeration has a U-shaped nexus with CO2 emissions for overall sample. The bidirectional causal relationship has also been confirmed between urbanization and CO2 emissions, between urban agglomeration and CO2 emissions, between financial development and CO2 emissions both in the long-run and short-run. On the other hand, unidirectional causality runs from economic growth, trade openness, and energy consumption to CO2 emissions in the long-run. The rising trend of urban agglomeration in metropolitan cities in South Asia is adversely affecting the environment. The current study has implications for policymakers and respective governments to adhere to more stringent urban planning.