A model to predict accurately pH evolution during alcoholic fermentation of must by Saccharomyces cerevisiae is proposed for the first time. The objective at least is to determine if the pH measurement could be used for predictive control. The inputs of the model are: the temperature, the concentrations in sugars, ethanol, nitrogen compounds. mineral elements (magnesium, calcium, potassium and sodium) and main organic acids (malic acid, citric acid, acetic acid, lactic acid, succinic acid). In order to avoid uncertainties coming from the possible precipitation. we studied this opportunity on a grape must without any tartaric acid, known as forming complexes with potassium and calcium during the fermentation. The model is based on thermodynamic equilibrium of electrolytic compounds in solution. The dissociation constants depend on the temperature and the alcoholic degree of the solution. The average activity coefficients are estimated by the Debbye-Huckel relation. A fictive diacid is introduced in the model to represent the unmeasured residual species. The molality of hydrogen ions and thus the pH are determined by solving a non-linear algebraic equations system consisted of mass balances, chemical equilibrium equations and electroneutrality principle. Simulation results showed a good capacity of the model to represent the pH evolution during fermentation. (C) 2007 Elsevier B.V. All rights reserved.