A variety of atomically dispersed transition-metal-anchored nitrogen-doped carbon (M−N−C) electrocatalysts have shown encouraging electrochemical CO 2 reduction reaction (CO 2 RR) performance, with the underlying fundamentals of central transition-metal atom determined CO 2 RR activity and selectivity yet remaining unclear. Herein, a universal impregnation-acid leaching method was exploited to synthesize various M− N−C (M: Fe, Co, Ni, and Cu) single-atom catalysts (SACs), which revealed d-orbital electronic configuration-dependent activity and selectivity toward CO 2 RR for CO production. Notably, Ni−N−C exhibits a very high CO Faradaic efficiency (FE) of 97% at −0.65 V versus RHE and above 90% CO selectivity in the potential range from −0.5 to −0.9 V versus RHE, much superior to other M−N−C (M: Fe, Co, and Cu). With the d-orbital electronic configurations of central metals in M−N−C SACs well elucidated by crystal-field theory, Dewar−Chatt−Duncanson (DCD) and differential charge density analysis reveal that the vacant outermost dorbital of Ni 2+ in a Ni−N−C SAC would benefit the electron transfer from the C atoms in CO 2 molecules to the Ni atoms and thuseffectively activate the surface-adsorbed CO 2 molecules. However, the outermost d-orbital of Fe 3+ , Co 2+ , and Cu 2+ occupied by unpaired electrons would weaken the electron-transfer process and then impede CO 2 activation. In situ spectral investigations demonstrate that the generation of *COOH intermediates is favored over Ni−N−C SAC at relatively low applied potentials, supporting its high CO 2 -to-CO conversion performance. Gibbs free energy difference analysis in the rate-limiting step in CO 2 RR and hydrogen evolution reaction (HER) reveals that CO 2 RR is thermodynamically favored for Ni−N−C SAC, explaining its superior CO 2 RR performance as compared to other SACs. This work presents a facile and general strategy to effectively modulate the CO 2to-CO selectivity from the perspective of electronic configuration of central metals in M−N−C SACs.