Though it is of great importance, the majority of predictive models tend not to incorporate water chemistry in their formulations. The ionic diffusion model which was developed for CaCO 3 , is based purely on crystallization, and is one of the few models that incorporates water chemistry. This model does not provide satisfactory predictions for CaSO 4 fouling. In this article, a new model is proposed for CaSO 4 fouling which takes into account the effect of both crystallization and particulate fouling and is capable of predicting the fouling resistance during the cleaning cycle as well as the fouling cycle. A removal term is incorporated into the model, as the occurrence of particulate fouling for CaSO 4 tends to weaken its crystalline structure and makes it more prone than CaCO 3 to removal. Properties of the electrolyte were evaluated using MINTEQA2 computer code, which is approved by the U.S. Environmental Protection Agency. In this model, particulate fouling is estimated using the physical mechanism for particle transport and adherence, crystallization is estimated by ionic diffusion, and the removal term is approximated using hydrodynamics of ow and deposit properties. The inclusion of both crystallization and removal terms incorporates the effects of both water chemistry and hydrodynamics of the ow and provides a relationship which not only can predict fouling but also can predict dissolution, by change of water quality and/or stopping the operation, or removal by shear stress. The proposed model was assessed using published experimental data. The results indicate that this model provides good predictions: the slope of predicted rates as a function of the experimental rates is 1.05. The experimental results, though limited in number, suggest that crystallization is not the main or only mechanism contributing to CaSO 4 fouling. Particulate fouling seems to be a major contributor. Further experimentation is in process to con rm the degree of particulate fouling and to substantiate or to modify the model accordingly.