Chemical oxidation by Fenton's reagent of a reactive azo dye (Procion Deep Red H-EXL gran) solution has been optimized making use of the experimental design methodology. The variables considered for the oxidative process optimization were the temperature and the initial concentrations of hydrogen peroxide and ferrous ion, for a dye concentration of 100 mg/L at pH 3.5, the latter being fixed after some preliminary runs. Experiments were carried out according to a central composite design approach. The methodology employed allowed to evaluate and identify the effects and interactions of the considered variables with statistical meaning in the process response, i.e., in the total organic carbon (TOC) reduction after 120 min of reaction. A quadratic model with good adherence to the experimental data in the domain analysed was developed, which was used to plot the response surface curves and to perform process optimization. It was concluded that temperature and ferrous ion concentration are the only variables that affect TOC removal, and due to the cross-interactions, the effect of each variable depends on the value of the other one, thus affecting positively or negatively the process response. (C) 2008 Elsevier B.V. All rights reserved.