To formulate efficient emission reduction policies, we explore the driving forces of CO2 emissions from the transport sector of four municipalities in China during 2000–2017 based on the temporal-spatial decomposition analysis model. The key factors causing the differences in CO2 emissions will be investigated in important years (i.e., 2000, 2005, 2010, and 2017). Two main results are found: (1) Energy intensity, as well as transportation intensity, is the dominant restraining factor confronting the transport sector’s CO2 emissions, while the income effect is reversed. (2) The results of the analysis further show what drives different emission levels (e.g., income effect, energy structure, and transportation structure) and four municipalities have greater potential in the reduction of CO2 emissions. We further discuss the root causes of the spatial-temporal differences in carbon emissions in the transport sector and analyse their future trends in detail. Based on this, we propose to pay attention to strengthen coordinated emission reduction, optimize the transportation mode, adjust the energy consumption structure, and select emission reduction strategies according to local conditions. In addition, emission reductions in the transportation sector are also affected by various factors such as social systems and transformation costs, and further research is needed in the future. •A spatial-temporal analysis is conducted to study China's transport emissions.•Bilateral–regional spatial decomposition model to decompose factors are used.•Intensities of energy and transport are crucial factors restraining CO2 emissions.•The income effect is the main contributor to increase CO2 emissions.•Municipalities have greater potential in the reduction of CO2 emissions.