Abstract: A method for designing the control law of a bi-input-multi-output system using the pole collocation method is presented.The pole collocation equations（Eqs 10 and ll）of a bi-input-multi-output system are derived in a different form from the general. In their derivation the eigenvalues and eigenvectors of the dynamic matrix[F] （see Eqn 1 ） of the open-loop system are used. Corresponding to every pole （or a pair of poles）, a real（or a complex）equation is deduced. A nature of Eqn 11 is verified （see Eqn 13）. Then Equation 11 can be weighted to minimize the mean square of the relative errors of the real and imaginary parts of poles.So each equation can be weighted separately to improve the accuracy of poles which have predominant effect on stability. But it is Iimpossible to weight the pole collocation equations in general form to improve the accuracy of a specific pole.The numerical results show that the accuracy of poles obtained by the present method is better than that obtained by the "rank one controller" under the same unperfect conditions.