This work investigates photocatalytic decompositions of HCHO in air over thin films of titanium oxide prepared on inside surfaces of glass tubes by the following two methods: Method I (the inside surface of a glass tube was covered with an aqueous H2O2 solution containing amorphous titanium oxide, followed by calcination at 500 degrees C to convert amorphous titanium oxide to anatase one) and Method II (the inside surface of a glass tube was covered with an aqueous H2O2 solution containing anatase titanium oxide, followed by heating at 100 degrees C). All the experiments were conducted in a photocatalytic reactor with a parallel array of nine light sources. It is found that there is no remarkable difference between the permeabilities of UV light through glass tube walls coated with the photocatalyst films, although the photocatalyst film prepared by Method I is more transparent than that by Method II. As previously observed with the photocatalyst film prepared by Method I, therefore, the photocatalyst film by Method II is found to increase the rate of decomposition of HCHO as a result of action of the UV light emitted from the light source in a given glass tube on the photocatalyst films that coat adjacent glass tubes. Observation by SEM reveals that the photocatalyst film prepared by Method I has a very smooth surface, whereas that by Method II has a rugged one. This is because anatase titanium oxide particles prepared by Method II possesses much larger crystal sizes and is therefore estimated to have a very large specific surface area. As a result, the photocatalytic reactor equipped with the photocatalyst glass tubes prepared by Method II is found to rapidly decompose HCHO especially in the region of its high concentration and also even when the humidity is very high. Kinetic analyses indicate that the photocatalytic decompositions of HCHO on the titanium oxide films prepared by the two methods obey Langmuir-Hinshelwood type kinetics. It is also found that the rate constant for Method II is 53 times larger than that for Method I, implying that the higher activity for Method II is due to its larger specific surface area. (C) 2008 Elsevier B.V. All rights reserved.