This study applies the panel smooth transition regression (PSTR) model to explore the impacts of real income, energy, and investment on the CO2-income nexus for 99 countries covering the period from 1971 to 2010. We find that in the full sample, as real income rises, CO2 emissions rapidly increase first, and then their increasing rate starts to slow down, while the environmental Kuznets curve (EKC) hypothesis for CO2 emissions is supported from the composite results of three income groups. Our results show that decreasing energy usage, improving energy efficiency, and enhancing clean energy usage could effectively ease the impacts of real income on CO2 emissions. Moreover, countries with different energy trade conditions and income levels have different CO2-income correlations, indicating that one size does not fit all. •We explore the non-linear CO2-income nexus.•The EKC hypothesis is found from the composite results of three income groups.•A small CO2-income nexus is found in countries with lower energy usage.•The CO2-income nexus in more energy efficiency and clean energy usage is smaller.•The energy trade conditions and income levels will influence the CO2-income nexus.