Osmotic dehydration of chestnut (Castanea sativa Mill.) was carried out using aqueous solutions of sodium chloride of different concentrations (17.0%, 22.0%, and 26.5% w/w) as osmotic media, and working at three temperatures (25, 35, and 45 degrees C). At different time intervals (up to 8 h) of contact with the osmotic media, total mass, solids gain, water loss, and moisture content were determined. The experimental data modelling was achieved employing an empirical model based on logistic dose response (LDR) curves and also with a diffusional model considering a constant value of the coefficient of diffusion and without shrinkage. At the same operating conditions, sodium chloride and moisture contents were obtained for different depths of the samples to determine composition profiles during osmotic dehydration. A quality parameter, as colour, was determined for each sample. The analysis of the results on osmotic dehydration of chestnut allow to conclude that the mass transfer rates are slower than in other products and the effect of the operation temperature is not relevant to enhance mass. The LDR model fits successfully the experimental data and also allows estimating the equilibrium contents of several characteristic parameters for osmotic dehydration process. The consideration of a diffusional model with a constant coefficient of diffusion gives satisfactory results, but taking into account the penetration of osmotic agent into the product it is not adequate at all (overcoat for long osmotic operation periods where the penetrations of dehydration front is more intensive). The analysis of WL/SG parameters reveal that the optimal values, in the range of experimental conditions assayed, were obtained when low temperatures (25 degrees C) and a 22.0% (w/w) sodium chloride concentration as osmotic media are employed. (c) 2005 Elsevier Ltd. All rights reserved.