Abstract: After the effect of multicollinearity on the existing MahalanobisTaguchi system (MTS) methods in multidimensional system optimization is analyzed, a new method called MoorePenrose (MP) generalized inverse matrix method of MTS is put forward. Then, comparison is made between the new method and the existing ones. It shows that the new method outperforms the existing ones in three ways: 1) it guarantees the existence of a solution; 2) the solution obtained is unique; and 3) the solution is a generic one. Thus, it is more effective than the existing ones in multidimensional system optimization. As a case study, the new method is used for the optimization of a bloodviscositydiagnose system and the results obtained in this paper are verified.