The electro-mechanical (EM) admittance signals acquired from piezoelectric transducers (PZT) surface bonded to the host structure are often used for structural health monitoring (SHM). However, it is well known that the method is susceptible to contamination from environmental and operational conditions. This paper introduces a co-integration method to remove dynamic load effects from electro-mechanical admittance data. The proposed method is based on the concept of co-integration that is partially built on the analysis of the non-stationary behavior of time series. Instead of directly using admittance signatures of PZT for damage detection, the analysis of the co-integrated residual obtained from the co-integration procedure of EM admittance responses and the resonant peaks frequency of the real part of admittance (conductance) are chosen as co-integrated variables. The experiments of aluminum beam bolt loosening identification, which is under dynamic stress, were carried out to verify the effectiveness of the proposed method. The results showed that the method can isolate damage-sensitive features from stress variations, so as to successfully detect the existence of damage.