Photo-discolouration of Orange II solutions at different concentrations was carried out in a 1l concentric reactor irradiated with a 254nm mercury lamp (125W) in the presence of Degussa TiO 2 P-25 dispersions. Experimentation involving discolouration of Orange II has been carried out using factorial design methodology for the simulation of the three critical variables affecting the dye discolouration: concentration of Orange II, pH and concentration of TiO 2 . A single polynomial function corresponding to optical absorbance of Orange II (Y) was constructed for the treatment of the experimental data obtained in the immersion photo-reactor. The function (Y) allowed the prediction of the residual discolouration after 20min photo-treatment. By following this model, a minimum of well-chosen experiments was performed to optimise Orange II photo-discolouration. The modelling takes into account that in the chemical system one reaction parameter determines the setting of another and vice-versa (interactive effects). The correlation factor between the experimental and predicted values by the polynomial expression for the discolouration of Orange II was better than 95%. In a typical photo-reactor run at pH 3, the concentration of TiO 2 could be decreased to values <0.5g/l without affecting the discolouration kinetics.