Nature uses organic molecules for light harvesting and photosynthesis but most man-made water splitting catalysts are inorganic semiconductors. Organic photocatalysts, while attractive because of their synthetic tunability, tend to have low quantum efficiencies for water splitting. Here we present a crystalline covalent organic framework (COF) based on a benzobis(benzothiophene sulfone) moiety that shows a much higher activity for photochemical hydrogen evolution than its amorphous or semi-crystalline counterparts. The COF is stable under long-term visible irradiation and shows steady photochemical hydrogen evolution with a sacrificial electron donor for at least fifty hours. We attribute its high quantum efficiency of FS-COF to its crystallinity, its strong visible light absorption, and its wettable, hydrophilic 3.2 nm mesopores. These pores allow the framework to be dye sensitized, leading to a further 61% enhancement in the hydrogen evolution rate up to 16.3 mmol g-1 h-1. The COF also retained its photocatalytic activity when cast as a thin film onto a support. Photocatalytic solar hydrogen production-or water splitting-offers an abundant clean energy source for the future. The use of dispersed, powdered photocatalysts or thin catalyst films is attractively simple, but so far, no catalyst satisfies the combined requirements of cost, stability and solar-to-hydrogen efficiency. Since the first report of TiO2 as a photocatalyst, 1 many inorganic semiconductors have been explored for water splitting, both in photoelectrochemical cells and as photocatalyst suspensions. 2-4 Recently, organic semiconductors have emerged as promising materials for photocatalytic hydrogen and oxygen evolution. 5-7 Poly(p-phenylene) was first reported as a photocatalyst for hydrogen evolution in 1985, 8,9 but its activity was poor and limited to the ultraviolet spectrum. Since then, more active organic materials have been reported as visible light photocatalysts for hydrogen production using sacrificial donors. This started with carbon nitrides 5,10 followed by poly(azomethine)s, 11 conjugated microporous polymers (CMPs), 6,12,13 linear conjugated polymers, 12,14-16 and covalent triazine-based frameworks (CTFs). 17-19 Carbon nitrides were further developed into hybrid systems that facilitate overall water splitting to produce both hydrogen and oxygen, for example by including metal co-catalysts. 20 CMPs were also claimed to exhibit overall photocatalytic water splitting. 21 However, while it is possible to tune semiconductor properties such as band gap by modular copolymerization strategies, 6 organic materials such as carbon nitrides, conjugated polymers and CTFs lack long-range order: they are amorphous or semi-crystalline. 17,22 This lack of order might limit the transport of photoactive charges to the catalyst surface. 23 More generally, it is challenging to construct atomistic structure-property relationships for materials where the three-dimensional architecture is poorly defined. Covalent organic frameworks (COFs) 24-26 are a cla...