A simple method for accurate identification of Baijiu was developed without qualitative and quantitative determination of volatile components. A total of 10 kinds of Baijiu with two flavor types were selected， and each of them has 2 samples from different blending batches. After directly headspace sample preparation， gas chromatography-ion mobility spectrometry (GC-IMS) was used to obtain the visualized fingerprints， and further extract the main features of Baijiu to establish Fisher linear discriminant models. A total of 134 volatile components were detected in the samples. The fingerprints can directly reflect the content difference of trace volatile components and the batch stability of blending. Under the premise of not fully analyzing and quantifying all components， stepwise regression analysis is used to optimize the three Fisher linear discriminant models which are respectively based on all 134 volatile components， 39 characteristic components selected by correlation analysis， and the 6 principal components obtained from principal component analysis. For the optimized correlation analysis model (the component features are reduced to 21) and the principal component analysis model， the classification accuracy rates and leave-one-out cross validation classification accuracy rates for the training set are 100%， and the discrimination accuracy rates for the test set are both 95%. The optimized Fisher linear discriminant model based on all the features has the best result， and 24 component features are extracted. The model's classification accuracy rate of the training set is 100%， the leave-one-out method cross-validation classification accuracy rate is 100%， and the prediction accuracy rate of the Baijiu samples in the test set reaches 100%. The GC-IMS is significant for the quality control， market supervision and the origin protection of Baijiu， and it can accurately identify Baijiu combining with pattern recognition.