A method is proposed for the design of constant-gain feedback-control laws for linear multivariable discrete-time systems which reduce trajectory sensitivity to small system parameter variations and ensure the closed-loop eigenvalues at the origin. The given quadratic index of performance including a sensitivity term is minimized in an average sense. An efficient computational procedure based on direct cost optimization using gradient type algorithm is also reported. An example is worked out to illustrate the proposed technique. © 1985 Taylor & Francis Group, LLC.