Abstract: Maize is susceptible to mildew infection,which affects food safety. Rapid determination of moldy maize is the prerequisite for controlling its hazards. This study intends to use the electronic nose(E-nose)technology based on gas sensor array to obtain the characteristic odor information of maize with different mildew degrees,and to establish a rapid detection method for mildew degree detection of maize. Irradiated sterilized maize was inoculated with 5 common harmful fungal strains of grains,and stored in 28 ℃ and 85% relative humidity for 15 days to serious mildew. On day 0,6,9,12 and 15,E-nose characteristic response signal of the sample’s odor information was collected,and qualitative and quantitative models towards maize mildew degree were established. The results showed that principal component analysis(PCA)can successfully distinguish maize samples with different mildew degrees. Linear discriminant analysis(LDA)has an average recognition rate of 93.3% for maize samples with different mildew degree infestation by single fungal,and 76.7% for all samples. The prediction coefficient(R_p²)of the partial least squares regression analysis(PLSR)model of the colony count in maize samples reached 0.777,the root mean square error of prediction set and the residual predictive deviation(RPD)was 0.981 log CFU/g and 2.12,respectively. The results showed that it is feasible to use the E-nose to quickly detect moldy maize. The next step is to expand the sample set size to improve the accuracy and reliability of the method.