Abstract Differentiated policies are key to improving the CO2 emissions reduction efficiency of cities, which are vital in mitigating climate change. The K-means cluster method and spatial logarithmic Divisia index decomposition method were used on the data of 279 cities in China to examine the impacts of local public expenditure on CO2 emissions in the context of socio-economic conditions. The results show emission differences in cities with similar socio-economic conditions. The impacts of the carbon intensity of local public expenditure and other public expenditures on the emission differences of city groups, which have different socio-economic conditions, were largest, followed by the local public expenditure scale and public environmental expenditure. The impacts of the proportions of public environmental expenditure and other expenditures were limited. Insights gained can provide feasible implications for Chinese cities, and enable policymakers to focus on the impact of different fiscal policies on CO2 emissions differences between cities.