The objective of this paper is to address a new method based on trend extraction for isolating faults in the nonstationary and nonlinear processes. Firstly, a concise review of the traditional methods for fault isolation based on Hotelling statistic are introduced, a rigorous analysis of their weaknesses, especially the smearing (coupling) phenomena, is provided, and the possible handling strategies are given. Secondly, a new contribution index for isolation is proposed based on the improved detection statistic, iT², and it's properties are analyzed. Finally, the effectiveness of the new contribution method is validated by a nonstationary and nonlinear numerical case. Also it is used for monitoring the satellite attitude control system. The results show that the new contribution method can cope with the smearing phenomena of the traditional contribution indexes.