Abstract: In order to improve the reliability of cigarette authenticity discrimination, a pattern recognition model based on t hypothesis test and SVM neural network was proposed, it discriminated counterfeit cigarettes from genuine ones according to their physical indexes. By randomly selecting 44 cigarette samples of a brand from a cigarette depot, the authenticity of samples was judged by the model. The results showed that the SVM neural network model based on three physical parameters(total ventilation rate, filter ventilation rate and hardness) could discriminate counterfeit cigarettes from genuine ones with an accuracy of 95%. It indicated that the method could be used to discriminate the authenticity of cigarette.