According to non-Gaussianity and randomness of the Pulse Signals, the bispectrum estimation is used to analyze the signals for the purpose of extracting the unusual information of the signals of-the drug abusers and educing the judgment of how to distinguish drug abusers from healthy persons. The phase of bispectrum carries the important feature information in signal processing. The feature vector of the average phase information based on the indirect nonparametric bispectrum estimation is extracted and used to analyze the Pulse Signals of 15 heroin addicts and 15 healthy persons. It is found that the average phase Ph of heroin addicts on a specified frequency region is generally lower than that of healthy persons. Using the average phase Ph as characteristic parameter, a critical parameter is determined that is used to classify heroin addicts and healthy persons. Thus, all of the 15 healthy addicts are identified. Only one heroin person is misjudged. Experiment results of classification show that the method is feasible and effective.