In this paper, multivariable quantitative feedback theory (QFT) techniques are used to synthesize a robust feedback system for an extractive distillation column with a vaporous sidestream. The simplified state space model of Gilles and Retzbach (1983) for an extractive distillation column exhibiting sharp temperature profiles is adopted here. Large uncertainty of +20% is assumed in three key matrix elements of this model. The feedback design problem is to find the controller and prefilter matrices such that the stated tracking specifications are met, over the entire range of model parametric uncertainty. The fourth MIMO QFT technique of Yaniv and Horowitz (1986) is used to solve the design problem. The design is verified through extensive simulations in both frequency and time domains, for several plants picked from the plant uncertainty set. In all cases, the results obtained are quite satisfactory.