Chinese liquors from different plants have unique flavors attributable to the use of various bacteria and fungi, raw materials, and production processes. Accurately identifying the flavor of Chinese liquors is not always possible through the subjective consciousness of a taster. A quartz crystal microbalance (QCM)-based electronic nose (e-nose) can perform this task because of its keen ability to imitate human senses. It does so by using a sensor array and a pattern-recognition system. In this paper, the behavior of a pattern-recognition system based on a random forest (RF) classifier is optimized by revising the number of decision trees and the number of variables in the decision trees of the RF. Raw data from the characteristics of Chinese liquors collected from the QCM-based e-nose were used by the RF classifier without processes of feature extraction and data pretreatment, which can reserve detailed information as much as possible. The prediction accuracies and computation times indicate a superior performance by the RF classifier over three other classifiers (linear discriminant analysis, backpropagation artificial neural network, and support vector machine). Taking both the application of the e-nose and the validation of the RF classifier into account, an available method is obtained to identify flavors of Chinese liquors.