Loop pairing is one of the major concerns when designing decentralized control systems for multivariable processes. Most existing pairing tools

such as the relative gain array (RGA) method

have shortcomings both in measuring interaction and in integrity issues. To evaluate the overall interaction among loops

we propose a statistics-based criterion via enumerating all possible combinations of loop statuses. Furthermore

we quantify the traditional concept of integrity to represent the extent of integrity of a decentralized control system. Thus

we propose that a pairing decision should be made by taking both factors into consideration. Two examples are provided to illustrate the effectiveness of the proposed criterion.