This paper presents an optimization based detuning parameter design method for tuning decentralized PI controllers for a multivariable process. The interactions, existing between the process input-output variables, make the control of MIMO processes a challenging problem. In this work, the MIMO control design problem is formulated into an optimization framework and the differential evolution approach is applied to achieve optimal performance. The framework provides for a suitable trade-off between performance and relative stability. In the proposed approach, the PI controller tuning problem is transformed into an equivalent optimal detuning parameter design problem. Thus dimensionality and the search space of the problem are reduced considerably. In order to verify the simplicity and applicability of the proposed method, closed loop simulations have been performed on various benchmark multivariable processes, including highly interacting and high dimensional MIMO systems. The effectiveness of the proposed approach is demonstrated by comparing performance against existing state of the art methods. © 2017 IEEE.