A simplified algorithm of prediction of water sorption isotherms for some foods was developed. This model is based on the composition of the main compounds of foods (glucose. fructose, sucrose, salt, protein, fibre and starch) and the influence of temperature was included (it was tested at 25 and 40 degrees C). Reported experimental data were employed as reference to validate the developed prediction model. Sorption isotherms for apple, apricot, banana, chestnut, loquat, quince, raisin, carrot, garlic, pepper, pumpkin, turnip, potato, bean, chickpea and lupine were predicted and compared to those reported in literature. The proposed model was able to predict the presence or absence of crossing between sorption isotherms at different temperatures for the same food. Using the prediction model could be calculated equilibrium moisture content with a determination coefficient (R(2)) of (>0.982), a mean relative error (E) of (<9.43%) and a standard error (E(RMS)) of (<0.042 kg (kg d.b.)(-1)). (C) 2009 Elsevier Ltd. All rights reserved.