Porous organic polymers (POPs) are promising precursors for developing high performance transition metal–nitrogen–carbon (M/N/C) catalysts for the oxygen reduction reaction (ORR). The rational design of POP precursors remain a great challenge, because of the elusive structural association between the sacrificial POPs and the final M/N/C catalysts. Based on covalent triazine frameworks (CTFs), we developed a series of S‐doped Fe/N/C catalysts by selecting six different aromatic nitriles as building blocks. A new mixed solvent of molten FeCl 3 and S was used for CTF polymerization, which benefited the formation of Fe–N x sites and made the subsequent pyrolysis process more convenient. Comprehensive study of these CTF‐derived catalysts showed that their ORR activities are not directly dependent on the theoretical N/C ratio of the building block, but closely correlated to the ratio of the nitrile group to benzene ring (N nitrile/N benzene) and geometries of the building blocks. The high ratios of N nitrile/N benzene are crucial for ORR activity of the final catalysts owing to the formation of more N‐doped micropores and Fe–N x sites in pyrolysis possess. The optimized catalyst shows high ORR performances in acid and superior ORR activity to the Pt/C catalysts under alkaline conditions.