The primary amine groups on the heptazine-based polymer melon, also known as graphitic carbon nitride (g-C3N4), can be replaced by urea groups using a two-step postsynthetic functionalization. Under simulated sunlight and optimum Pt loading, this urea-functionalized carbon nitride has one of the highest activities among organic and polymeric photocatalysts for hydrogen evolution with methanol as sacrificial donor, reaching an apparent quantum efficiency of 18% and nearly 30 times the hydrogen evolution rate compared to the nonfunctionalized counterpart. In the absence of Pt, the urea-derivatized material evolves hydrogen at a rate over four times that of the nonfunctionalized one. Since defects are conventionally accepted to be the active sites in graphitic carbon nitride for photocatalysis, the work here is a demonstrated example of defect engineering, where the catalytically relevant defect is inserted rationally for improving the intrinsic, rather than extrinsic, photocatalytic performance. Furthermore, the work provides a retrodictive explanation for the general observation that g-C3N4 prepared from urea performs better than those prepared from dicyandiamide and melamine. In-depth analyses of the spent photocatalysts and computational modeling suggest that inserting the urea group causes a metal-support interaction with the Pt cocatalyst, thus facilitating interfacial charge transfer to the hydrogen evolving centers.