In process monitoring, several diagnosis methods have been used for fault diagnosis. These methods have been developed from different backgrounds and considerations. In this paper, five existing diagnosis methods are analyzed and generalized. It is shown that they can be unified into three general methods, making the original diagnosis methods special cases of the general ones. Also, a new form of relative contributions is proposed. An analysis of the diagnosability shows that some diagnosis methods do not guarantee correct diagnosis even for simple sensor faults with large magnitudes. For faults with modest fault magnitudes, Monte Carlo simulation is applied to compare the performance of the diagnosis methods. (C) 2010 Elsevier Ltd. All rights reserved.