For several decades, the main body of research in photoelectrochemical (PEC) hydrogen production has centered on a small number of semiconductor materials classes, including stable but inefficient metal-oxides, as well as some more efficient materials such as III-V compounds which suffer from high cost and poor stability. While demonstrating some limited success in meeting the rigorous PEC demands in terms of bandgap, optical absorption, band-edge alignment, surface energetics, surface kinetics, stability, manufacturability and cost, none of the 'traditional' PEC semiconductors are adequate for application in water-splitting devices with high performance (greater than 15% solar-to-hydrogen conversion) and long durability (greater than 200 h life). As a result, it is widely held that new semiconductor classes and configurations need to be identified and developed specifically for practical implementation of solar water-splitting. Examples include ternary and quaternary metal-oxide compounds, as well as non-oxide semiconductor materials, such as silicon-carbide and the copper-chalcopyrites. This paper describes recent progress at the University of Hawaii to develop improved semiconductor absorbers and interfaces for solar photoelectrolysis based on polycrystalline tungsten trioxide and polycrystalline copper-gallium-diselenide. Specific advantages and disadvantages of both materials classes in terms of meeting long-term PEC hydrogen production goals are detailed. Copyright (C) 2010 John Wiley & Sons, Ltd.