Abstract: Abstract: Egg has become an indispensable food product in our daily life, which contains large amounts of protein and balanced nutrition for body development. One of the major concerns of the egg industry is to determine the freshness of eggs in an efficient and nondestructive way. Quartz crystal microbalance (QCM) gas sensors have received more attention in recent years, which provide a simple and nondestructive method to test egg samples by sensing the volatiles released by whole eggs. By measuring the resonance frequency shift caused by the adsorption process, nanogram-level mass change can be detected, and the gas samples can be easily estimated. QCM technique has been widely used in the field of gas detection because of its high sensitivity, low operating temperature and facile operation. In this study, a QCM sensor array with 4 different surface modified QCM sensors, i.e., multi-walled carbon nanotubes (CNTs), graphene, nanostructured copper oxide (CuO) and polyaniline nanocomposite (PANI), was fabricated by dip coating method. The frequency shift of sensors in the process of absorption and desorption was monitored by a self-made frequency measurement system. The morphologies of sensitive materials were characterized by field-emission scanning electron microscope (FE-SEM), and the number of sensitive materials coating layers was firstly optimized. To balance both the sensitivity and the response time, 4 layers of CNTs, 4 layers of graphene, 5 layers of CuO and 5 layers of PANI, were determined to be deposited on the surface of the QCM sensors, which together formed a QCM sensor array, and were applied in further study. Then the gas sensing properties of the 4 sensors were tested by ethanol vapor, which is one of the volatiles existing in stale eggs. The result demonstrated that the frequency shift of the QCM sensors was stable when responding to the same concentration of ethanol, and with the increase of ethanol concentration, the frequency shifts of the QCM sensors increased gradually and showed a linear relationship with the vapor concentration, which exhibited promising repeatability, reversibility and good sensitivity. The sensitivity of the 4 sensors was 2.31 (CuO), 2.05 (CNTs), 1.70 (PANI) and 1.37 Hz/(mg/kg) (Graphene), respectively. Moreover, the long-term stability of the 4 sensors was all above 85% during 30 days' test. The experiment results validated the reliability of the fabricated QCM sensor array so that it could be used in the evaluation of eggs. In further application, the sensor array was used to detect eggs with different shelf lives. The responding curve of frequency shift in the sensor was extracted as the feature for statistics analysis. Principal component analysis (PCA) and linear discriminant analysis (LDA) were employed to classify eggs with different shelf lives. The first 2 PCs of PCA explained 84.88% of total variations and could represent the original data. The discrimination results of LDA outperformed the results of PCA, which separated all classes of data points completely in the plot. LDA method exhibited a good classification accuracy with 100% in the original data and 98.8% in cross-validation procedure. As for the prediction of shelf life of egg, partial least square regression (PLSR) was used to establish a regression model and the coefficient of determination (R2) and the root mean square error (RMSE) of the regression model were employed as the criteria of the regression model. The kernel principal component analysis (KPCA) was used to solve the nonlinear relation in original dataset. The PLSR regression model showed a good prediction performance, in which the R2 increased from 0.847 4 to 0.954 7 and the RMSE decreased from 3.026 8 to 1.666 1 d after KPCA was introduced. It could be concluded that the fabricated QCM sensor array is effective for the evaluation of eggs with different shelf lives, offering a sensitive, nondestructive and simple method to evaluate the shelf life of eggs.