One of the most popular techniques for altering the response characteristics of a control system is the application of linear state variable feedback .The fact of using state feedback to assign the closed loop system self-conjugate set of eigenvalues, provided that the open loop system is controllable, is a well known and commonly used technique. For single-input systems, the state feedback gain is uniquely determined by the desired pole pattern, but for multiinput systems, there is a lot of freedom in choosing the parameters of the feedback gain matrix, which can assign a specified eigenvalue spectrum. There are few systematic ways of finding a unique gain feedback matrix, such as minimizing a quadratic performance index in which case the gain will be obtained by solving a Riccatti equation, or assigning both the eigenvalues and their eigenvectors .The main contribution of this paper is the elaboration of a new procedure for obtaining a unique state feedback gain matrix, which places the closed loop system poles to the desired locations and meets the following criteria: 1_Achieving the best possible time response characteristics, 2_Yielding a system with small feedback gains, 3_ Yielding a system with a good robustness.