A new algorithm for non-square internal model control systems based on the pole placement approach has been developed. The algorithm can handle any n-input and m-output minimum- or non-minimum-phase systems. The process factorization and inversion are combined into the derivation of the control law so that these procedures can be avoided. A tuning gain is introduced in the algorithm to improve the performance, and can be determined in a stable region by an optimization technique. The offset is eliminated through the integral action of the errors in the algorithm. Moreover, the overall characteristic behaviour of the control system consists of the feedforward and feedback control behaviour so that it has both tracking and disturbance rejection abilities. Comparisons with classical internal model control (IMC) and simplified model predictive control design for m × m square systems are also discussed.