Abstract: Multivariate statistical process monitoring (MSPM) is an efficient data-driven fault detection and diagnosis approach forcomplex industrial processes. Partial least squares or projection tolatent structures (PLS) is one of the latent projection structuresused in MSPM, which uses process data X and quality data Y together. In this paper, we discuss a new fault diagnosis approachbased on total projection to latent structures (T-PLS). Four kindsof monitoring statistics are used in T-PLS, and a new definition ofvariable contributions to T2 of PLS is proposed. Then, definitions of variable contributions to all statistics are derivedto identify the faults. Control limits for contribution plots arecalculated to identify whether a variable is in abnormal situationor not. Further, the proposed method separates the identifiedvariables into faulty variables related to Y and unrelated to Y more clearly than conventional method. A case study on TennesseeEastman process (TEP) indicates the efficiency of the proposed approach.