The rate of hydrogen evolution from a photocatalytic process depends not only on the activity of a photocatalyst, but also on photoreactor design. Ideally, a photoreactor should be able to absorb the incident light, promoting photocatalytic reactions in an effective manner with minimal photonic losses. There are numerous technical challenges and cost related issues when designing a large-scale photoreactor for hydrogen production. Active stirring of the photocatalyst slurry within a photoreactor is not practical in large-scale applications due to cost related issues. Rather, the design should allow facile self-mixing of the flow field within the photoreactor. In this paper two types of photocatalytic reactor configurations are studied: a batch type design and another involving passive self-mixing of the photolyte. Results show that energy loss from a properly designed photoreactor is mainly due to reflection losses from the photoreactor window. We describe the interplay between the reaction and the photoreactor design parameters as well as effects on the rate of hydrogen evolution. We found that a passive self-mixing of the photolyte is possible. Furthermore, the use of certain engineering polymer films as photoreactor window materials has the potential for substantial cost savings in large-scale applications, with minimal reduction of photon energy utilization efficiency. Eight window materials were tested and the results indicate that Aclar (TM) polymer film used as the photoreactor window provides a substantial cost saving over other engineering polymers, especially with respect to fused silica glass at modest hydrogen evolution rates. Published by Elsevier Ltd.