In this work, published experimental result data of the pulping of tagasaste (Chamaecytisus proliferus L.F.) with soda and anthraquinone (AQ) have been used to develop a model using a neural network. The paper presents the development of a model with a neural network to predict the effects that the operational variables of the pulping reactor (temperature, soda concentration, AQ concentration, time and liquid/solid ratio) have on the properties of the paper sheets of the obtained pulp (brightness, traction index, burst index and tear index). Using a factorial experimental design, the results obtained with the neural network model are compared with those obtained from a polynomial model. The neural network model shows a higher prediction precision that the polynomial model. (C) 2008 Elsevier Ltd. All rights reserved.