The objective of this study was to define, from literature data, the most relevant water sorption isotherm to use in simulation models for cassava root processing. The isotherm should be as global and generic as possible. Compiling isotherms from the literature showed that the variety of cassava and the level of processing generally have a negligible or unclear influence on the sorption behavior. Moreover, no significant hysteresis between adsorption and desorption is observed. For the purpose of simulation, measuring the sorption properties specific to each batch of product appears to be irrelevant. Actually, for this purpose, global properties, representing the average product behavior are preferred to propose equipment pre-design available for a large range of products. Following the wisdom of crowds theory, it is assumed that the mean value of sorption measurements with uncertainties on products from various origins from a dataset that is sufficiently large will represent the behavior of an average product with an accuracy sufficient for the purpose of simulation. A global, generic model (Eq. (1)) was obtained by fitting all the ad- and desorption data reviewed to a temperature-dependent model. A model of the heat of sorption was deduced from this global model (Eq. (2)). Both models are valid in a 25-90° C temperature range and a 0.05-0.93 water activity range.