This work reports the use of a hybrid modeling approach to represent an industrial cooking process of chewy candy. The target process comprises an evaporation and a vacuum drying step and hence the main goal is to estimate the final moisture of the product. The hybrid approach presented here is constituted by a phenomenological model to the evaporation step and an artificial neural network to model the vacuum drying step. As reliability and robustness of the phenomenological model is strongly dictated by availability of experimental information about the raw material to be processed, experimental data of boiling point elevation of chewy candy syrup were measured using an ebulliometric method. The approach proposed here was correlated with data from steady state and transient conditions measured in the industrial process environment. Results obtained in this work demonstrate that the approach adopted was able to represent very well the chewy candy moisture and may be considered an interesting alternative to model this kind of system. (C) 2008 Published by Elsevier Ltd.