Abstract: In this paper,a multivariable system is described by a time-invariant model obtained from real system step responses and the uncertainty is represented by the difference between model step responses and real system ones.A state feedback law is designed such that the closed-loop system consisting of model and controller has the prescribed pole locations and at the same time the largest stability margin is searched via optimization techniques for the remained free gains in state feedback matrix.This method is applied to papermachine control and satisfactary performance robustness results.