The viscosity of multimodal suspensions was studied by expanding the unimodal suspension model with the assumption that transported momentum by particle collision is proportional to the momentum difference (Ookawara and Ogawa, 2000). It was shown that the derived multimodal model could predict the larger viscosity of a mixture than the viscosity of each suspension. The prediction of this tendency was impossible by previous models that assumed the independent contribution of each component (Farris, 1968). It was confirmed that, moreover, the theoretical equation could predict the various blend effects, viz., not only the average viscosity of two suspensions, but also whether the blend would have larger or smaller viscosity than the two suspension viscosities. This finding suggests that the interaction between different size particles is dominant in determining the viscosity of a blend suspension. Further, the relation between statistical index of particle size distribution and suspension viscosity was derived based on the newly developed model. Finally, the usefulness of the derived equation was discussed for the estimation of classification effect and viscosity fluctuation in industry.