This paper points out the importance of the multitemperature fitting procedure in description of water sorption on foodstuffs. The data tabulated in literature (water sorption at different temperatures on: chickpea seeds, lentil seeds, potato and on green peppers) were described applying the BET, GAB and recently proposed GDW models. Our results explain total failure of the first model in description of multitemperature data and the similarities between the GAB and GDW are shown. Finally the general mechanism of water sorption on foodstuffs is proposed. This mechanism can be of the GAB or GDW type, depending on the arrangement and features of the primary water sorption sites. If the geometrical constraints for creation of the BET - like type clusters do not occur on surface, and if each from primarily sorbed water molecules convert only into one secondary surface site, one can say that the mechanism follows the GAB scenario (as for example in the case of lentil seeds). Contrary, in the case of rough or porous surfaces, where there are the geometric constraints for creation of secondary sites (for example sorption on chickpea seeds), and/or where one primary site produces more than one secondary site (potato and green peppers), the mechanism of water sorption is of the GDW type. (C) 2007 Elsevier Ltd. All rights reserved.