The mechanism of the photoelectrochemical oxidation of methanol and salicylic acid on anatase film electrodes was studied as a function of the applied potential and pollutant concentration at pH 3. The dependencies of the steady state photocurrents on substrate concentration reflect the type of surface interaction: weak in the case of methanol, that leads to a simple saturation curve, and strong in the case of salicylate, that shows a steady state photocurrent peaking at intermediate concentrations. At 0.6 V vs. SCE the oxidation rate is largely enhanced as compared to open circuit conditions (E-oc = -0.3 V). Even under nitrogen, the reaction proceeds at an appreciable rate, and the ratio of circulated charge to the number of oxidized salicylate ions approaches 28 electrons per mol at low salicylate concentration: oxidized salicylate mineralizes almost totally, and the intermediates are rapidly destroyed. At higher substrate concentrations, the ratio decreases, and uv spectral evidence suggests the formation of some undefined oxidation products. Under oxygen at 0.6 V, the radicals generated in the initial photoelectrochemical step are mostly oxidized by O-2, increasing the amount of salicylate destroyed for a given total circulated charge; at sufficiently high substrate concentration, the above ratio decreases to values below 4. No evidence of the presence of traces of partially oxidized molecules is found. Adequate control of the experimental conditions permits therefore to achieve substantially increased efficiencies of salicylate destruction per absorbed photon, and the build up of uncontrolled intermediates can be prevented. The results are discussed in terms of the oxidation length Y, defined as the number of oxidation steps that are triggered by a single hole transfer event, and of the oxidation efficiency epsilon, defined as the ratio of the oxidation length to the maximum possible oxidation length (the length achieved when one hole transfer suffices to trigger total mineralization).