Abstract As the capital of China, Beijing is regarded as a major metropolis in the world. Study of the variation in temporal CO2 emissions generated by the driving forces in Beijing can provide guidance for policy decisions on CO2 emissions mitigation in global metropolises. Based on input–output structural decomposition analysis (IO-SDA), we analysed the driving forces for the increment in CO2 emissions in Beijing from both production and final demand perspectives during 1997–2010. According to our results, the CO2 emission growth in Beijing is driven mainly by production structure change and population growth, partly offset by CO2 emission intensity reduction as well as the decline in per capita final demand volume during the study period. Final demand structure change has a limited effect on the change in the CO2 emissions in Beijing. From the final demand perspective, urban trades, urban residential consumption, government consumption and fixed capital formation are mainly responsible for the booming emissions. This study showed how the “top-down” IO-SDA methodology was implemented on a city scale. Policy implications from this study would be helpful for addressing CO2 emissions mitigation in global capital cities and metropolises.