The　goal　of　this　study　is　to　explore　the　causal　relationship　among　economic　growth,　economic　complexity　and　CO2　emissions　by　using　panel　data　of　95　countries　for　the　period　1996-2015.　A　novel　panel　Granger　approach　proposed　by　Juodis　et　al.　(2021)　is　adopted.　Under　this　approach,　we　can　explore　the　Granger　causality　in　homogeneous　or　heterogeneous　panels.　To　uncover　the　heterogeneous　causal　effects　at　different　income　levels,　this　study　further　divide　the　sample　into　three　groups　according　to　their　annual　income　levels.　Empirical　results　show　that　there　are　bi-directional　causalities　among　economic　growth,　economic　complexity　and　CO2　emissions　for　all　groups.　However,　the　magnitudes　of　the　effects　are　heterogeneous　for　different　groups.　As　to　low-income　countries,　economic　complexity　is　positive　and　significant　for　CO2　emissions,　while　CO2　emissions　are　negative　for　economic　complexity.　Furthermore,　there　is　a　positive　interaction　between　economic　complexity　and　C(O)2　emissions　for　middle-income　countries.　Regarding　high-income　countries,　however,　increasing　economic　complexity　might　effectively　reduce　CO2　emissions,　and　CO2　emissions　can　significantly　increase　economic　complexity.　Additionally,　economic　complexity　will　prominently　decrease　GDP　in　low-income　countries.　These　findings　are　robust　to　different　economic　complexity　indexes.　Our　results　suggest　that　both　developing　and　developed　countries　should　set　a　reasonable　CO2　emissions　target,　and　on　this　basis,　maintain　a　good　balance　between　economic　complexity　and　CO2　emissions.　(C)　2021　Economic　Society　of　Australia,　Queensland.　Published　by　Elsevier　B.V.　All　rights　reserved.