Transmembrane(TM) protein plays an important role in the life activity of the cells, and the prediction of transmembrane helical segments (TMHs) is an important subject in the bioinformatics research. Thus far, several prediction methods have been report ed, but there are some deficiencies in prediction accuracy and adaptability in these methods. In this paper, a method based on discrete wavelet transform (DWT) was developed to predict the TMHs. Two sets of test data sets containing total 60 protein sequences were utilized to access the effect of the method. Compared with the prediction results of TMHMM2.0 and MEMSAT, the obtained results indicate that the presented method has high prediction accuracy.