Assessing the effectiveness of control systems and limiting the incidence of fault events are main emphases of studies on performance assessment of control systems. In single-input/single-output (SISO) systems, the performance bounds are easily estimated from the minimum variance of a single variables however, in multi-input/multi-output (MIMO) systems, the benchmark is computed from traces of multivariable covariance matrices. This MIMO method of performance assessment sometimes can cause false detection because it does not consider the correlation among variables. In this article, a new method, called PCA-ARMA-based performance assessment, is proposed. It is an integrated framework of multivariable principal component analysis (PCA) and the autoregressive moving average (ARMA) filter. It can build up the PCA-based minimum-variance performance bound of the MIMO feedback control system under minimum-variance control. In this integrated framework, the difference between the process outputs and the desired set points is projected onto the PCA space to determine the scores and residuals. An ARMA filter is selected to eliminate the autocorrelation of the scores. The independent scores and residuals are used to compute two statistics, which serve as performance indices in monitoring the progress of operation. Because MV control of processes is at times impractical or even infeasible, the integrated PCA and ARMA method, which is used to design achievable-minimum-variance (AMV) control, is proposed. AMV control is proposed to tune the controller parameters and minimize the weighted sum of variances of the various scores. Again, the multivariable PCA and the ARMA filter build up the AMV performance bound of the MIMO feedback control system under AMV control. The integrated method can result in better monitor indices. The performance of the proposed AMV method illustrated through a simulation case and a quadruple-tank experiment. The case studies show that the proposed method yields better monitoring results than the conventional approaches in terms of sensitivity and response time to system degradation.