Photocatalytic ozonation (light/O-3/photocatalyst), an independent advanced oxidation process (AOP) proposed in 1996, has demonstrated over the past two decades its robust oxidation capacity and potential for practical wastewater treatment using sunlight and air (source of ozone). However, its development is restricted by two main issues: (i) a lack of breakthrough catalysts working under visible light (42-43% of sunlight in energy) as well as ambiguous property-activity relationships and (ii) unclear fundamental reasons underlying its high yield of hydroxyl radicals ((OH)-O-center dot). In this Account, we summarize our substantial contributions to solving these issues, including (i) new-generation graphitic carbon nitride (g-C3N4) catalysts with excellent performance for photocatalytic ozonation under visible light, (ii) mechanisms of charge carrier transfer and reactive oxygen species (ROS) evolution, (iii) property-activity relationships, and (iv) chemical and working stabilities of g-C3N4 catalysts. On this basis, the principles/directions for future catalyst design/optimization are discussed, and a new concept of integrating solar photocatalytic ozonation with catalytic ozonation in one plant for continuous treatment of wastewater regardless of sunlight availability is proposed. The story starts from our finding that bulk/nanosheet/nanoporous g-C3N4 triggers a strong synergy between visible light (vis) and ozone, causing efficient mineralization of a wide variety of organic pollutants. Taking bulk g-C3N4 as an example, photocatalytic ozonation (vis/O-3/g-C3N4) causes the mineralization of oxalic acid (a model pollutant) at a rate 95.8 times higher than the sum of photocatalytic oxidation (vis/O-2/g-C3N4) and ozonation. To unravel this synergism, we developed a method based on in situ electron paramagnetic resonance (EPR) spectroscopy coupled with an online spin trapping technique for monitoring under realistic aqueous conditions the generation and transfer of photoinduced charge carriers and their reaction with dissolved O-3/O-2 to form ROS. The presence of only 2.1 mol % O-3 in the inlet O-2 gas stream can trap 1-2 times more conduction band electrons than pure O-2 and shifts the reaction pathway from inefficient three-electron reduction of O-2 (O-2 -> O-center dot(2)- -> HO2 center dot -> H2O2 -> (OH)-O-center dot) to more efficient one-electron reduction of O-3 (O-3 -> O-center dot(3) -> HO3 -> (OH)-O-center dot), thereby increasing the yield of 'OH by a factor of 17. Next, we confirmed band structure as a decisive factor for catalytic performance and established a new concept for resolving this relationship, involving "the number of reactive charge carriers". An optimum balance between the number and reducing ability of photoinduced electrons, which depends on the interplay between the band gap and the conduction band edge potential, is a key property for highly active g-C3N4 catalysts. Furthermore, we demonstrated that g-C3N4 is chemically stable toward O-3 and O-center dot(2)- but that (OH)-O-center dot can tear and oxidize its heptazine units to form cyameluric acid and further release nitrates into the aqueous environment. Fortunately, (OH)-O-center dot usually attacks organic pollutants in wastewater in preference to g-C3N4, thus preserving the working stability of gC(3)N(4) and the steady operation of photocatalytic ozonation. This AOP, which serves as an in situ (OH)-O-center dot manufacturer, would be of interest to a broad chemistry world since (OH)-O-center dot radicals are active species not only for environmental applications but also for organic synthesis, polymerization, zeolite synthesis, and protein footprinting.