The impacts of urbanization on carbon emissions have attracted widespread attention for a long time. Quantitative research on the impacts is of great significance for formulating carbon reduction policy. Based on the dynamic panel autoregressive distribution lag (ARDL) model, we systematically study the general and heterogeneous long-run equilibrium relationships, short-run dynamic relationships, impact mechanism and lag effect between urbanization and three carbon emission dimensions in OECD high-income countries during long period. The main empirical results indicate that developed countries tend to have the same negative impacts of urbanization on carbon emissions, although there are differences in the endowments of different countries. The impact is relatively weak, for each percentage point increase in urbanization rate, CO2 emissions per capita decrease by 0.015%, total CO2 emissions decrease by 0.012%, and CO2 emission intensity decrease by 0.009%. All member countries have achieved the decoupling of urbanization and carbon emissions. Urbanization affects carbon emissions by affecting economic growth, energy efficiency, and final energy consumption structure. This paper further reveals the multi-level impacts of urbanization on carbon emissions and provides policy implications of achieving carbon reduction through urbanization's agglomeration effect for government decision makers. •Estimate the impact of urbanization on carbon emissions based on panel ARDL model.•Analyze general & heterogeneous long-, short-run relationship and impact mechanism.•Urbanization decreases carbon emissions but the impact is weak in OECD countries.•Developed economies have achieved the decoupling of urbanization and CO2 emissions.•Promote urbanization process and exert its scale effect to reduce carbon emissions.