For better quality control of paper machines, variations in the machine direction (MD), cross-machine direction (CD) and their inherent interactions should be minimized. In this paper, a dynamic MD-CD interaction model is developed by relating the effect of MD dry basis weight to the CD profile. Based on this model, a combined MD-CD generalized predictive control strategy is proposed to handle the strong MD-CD interaction. A set of industrial data was used to identify the interaction model. Results from closed-loop simulation of the interaction model under the proposed control strategy show that a significant improvement in CD control during a grade change can be achieved.