R esearchers are intensively investigating photochemical water splitting as a means of converting solar to chemical energy in the form of fuels. Hydrogen is a key solar fuel because it can be used directly in combustion engines or fuel cells, or combined catalytically with CO 2 to make carbon containing fuels. Different approaches to solar water splitting include semiconductor particles as photocatalysts and photoelectrodes, molecular donor-acceptor systems linked to catalysts for hydrogen and oxygen evolution, and photovoltaic cells coupled directly or indirectly to electrocatalysts.Despite several decades of research, solar hydrogen generation is efficient only in systems that use expensive photovoltaic cells to power water electrolysis. Direct photocatalytic water splitting is a challenging problem because the reaction is thermodynamically uphill. Light absorption results in the formation of energetic charge-separated states in both molecular donor-acceptor systems and semiconductor particles. Unfortunately, energetically favorable charge recombination reactions tend to be much faster than the slow multielectron processes of water oxidation and reduction. Consequently, visible light water splitting has only recently been achieved in semiconductor-based photocatalytic systems and remains an inefficient process.This Account describes our approach to two problems in solar water splitting: the organization of molecules into assemblies that promote long-lived charge separation, and catalysis of the electrolysis reactions, in particular the four-electron oxidation of water. The building blocks of our artificial photosynthetic systems are wide band gap semiconductor particles, photosensitizer and electron relay molecules, and nanoparticle catalysts. We intercalate layered metal oxide semiconductors with metal nanoparticles. These intercalation compounds, when sensitized with [Ru(bpy) 3 ] 2+ derivatives, catalyze the photoproduction of hydrogen from sacrificial electron donors (EDTA 2-) or non-sacrificial donors (I -). Through exfoliation of layered metal oxide semiconductors, we construct multilayer electron donor-acceptor thin films or sensitized colloids in which individual nanosheets mediate light-driven electron transfer reactions. When sensitizer molecules are "wired" to IrO 2 · nH 2 O nanoparticles, a dye-sensitized TiO 2 electrode becomes the photoanode of a water-splitting photoelectrochemical cell.Although this system is an interesting proof-of-concept, the performance of these cells is still poor (∼1% quantum yield) and the dye photodegrades rapidly. We can understand the quantum efficiency and degradation in terms of competing kinetic pathways for water oxidation, back electron transfer, and decomposition of the oxidized dye molecules. Laser flash photolysis experiments allow us to measure these competing rates and, in principle, to improve the performance of the cell by changing the architecture of the electron transfer chain.