This paper reports measurements of photocatalytic oxidation of 2-propanol by dispersions of three different types (a high area rutile, a surface treated rutile, and, the mainly anatase, Degussa P25) of titanium dioxide in 2-propanol. For a fixed loading of any one type of TiO2, the particle dispersion was deliberately modified by milling the suspension to break down the clusters of individual crystals which are usually present in dispersions of all inorganic powders. Sedimentation analysis demonstrated that, as predicted, milling decreased the size of the TiO2 particles and this led to corresponding changes in the optical properties of their dispersions. Measurements of 2-propanol oxidation demonstrated that, for all three forms of titania, milling-times of 30 min led to an 80% decrease in the rate of acetone formation. We propose that the measured decrease is a consequence of the demonstrated increase in UV absorption associated with the reduction of particle size caused by milling. This increased absorption causes the UV flux incident on the catalyst slurry to be absorbed within a shorter path length, i.e., is absorbed by fewer crystals, say 1/F of the original number. Therefore, although the total number of charge-carriers generated by a fixed UV flux is constant, more (F times as many) charge-carriers/crystal are generated within a smaller number (1/F) of crystals. However, since electron-hole recombination statistics are bimolecular, if the photon absorption per crystal increases by some factor, F, the number of hydroxyl radicals generated in each crystal increases not by a factor F, but only by F-0.5. Because the number of hydroxyl radicals generated, the number of radicals/crystal multiplied by the number of crystals, varies as F-0.5 x (1/F) and the total hydroxyl radical generation rate varies as F-0.5. Since experimental measurements reflect the total rate of hydroxyl radical generation, the acetone production rate decreases as the UV absorption increases. This explanation of these counter-intuitive changes is relevant to all reactions carried out using semiconductor catalyst particles dispersed in a liquid. It is of special relevance to the many comparisons of the photoactivity of different types of TiO2.