The technology of dividing wall columns can offer enormous energy savings compared to common distillation columns and configurations. The technology of model predictive control is also advantageous since such a controller minimizes the future deviation of the predicted controlled variable from the reference point. The practical application of model predictive controllers for dividing wall columns is still limited due to limited experience with high interactions among the process variables. The scope of this work is the development and analysis of a method for the design of model predictive controllers for dividing wall columns. An experimental investigation verifies the practicability of the applied approach. The methods generated are transferable to other applications. Thus, the industrial acceptance of model predictive controllers for dividing wall columns is enhanced.