The photocatalytic splitting of water into molecular hydrogen and molecular oxygen with sunlight is the dream reaction for solar energy conversion. For decades, transition metal oxide semiconductors and supramolecular organometallic structures have been extensively explored as photocatalysts for solar water splitting. More recently, polymeric materials consisting of triazine or heptazine building blocks have attracted considerable attention as hydrogen‐evolution photocatalysts. The mechanism of hydrogen evolution with these materials is discussed throughout the current literature in terms of the familiar concepts developed for photoelectrochemical water splitting with semiconductors since the 1970s. We discuss in this Minireview an alternative mechanistic paradigm for photoinduced water splitting with carbon nitrides, which focusses on the specific features of the photochemistry of aromatic N‐heterocycles in aqueous environments. Using ab initio electronic‐structure calculations, it is shown that a water molecule which is hydrogen‐bonded to an N‐heterocycle can be decomposed into hydrogen and hydroxyl radicals by two simple sequential photochemical reactions. This concept is illustrated by the discussion of excited‐state reaction paths and their energy profiles for hydrogen‐bonded complexes of pyridine, triazine and heptazine with a water molecule. It is shown that the excited‐state hydrogen‐transfer and hydrogen‐detachment reactions are essentially barrierless, in sharp contrast to water oxidation in the electronic ground state, where high barriers prevail. We also discuss in some detail the products of possible reactions of the highly reactive photogenerated hydroxyl radicals with the chromophores. We hypothesize that the challenge of efficient solar hydrogen generation with carbon nitride materials is less the decomposition of water as such, but rather the controlled recombination of the photogenerated radicals to the closed‐shell products H2 and H2O2.