Alternative splicing is the main mechanism expanding transcript diversity. Cassette exon is an important alternative splicing form that is very similar to constitutive exon in sequence features. Previous studies which based on expressed sequence tags (ESTs) and evolutionary conservation information have identified many alternative splicing events. In this paper, we construct a classifier to identify the human cassette exons based on Support vector machine (SVM) which only make use of sequence information. It can achieve the accuracy of 68.12%. Especially, the classifier can achieve 76.54% when considering the splicing frequency. The results show that the sensitivity and specificity of this method are higher than those recently reported on the same dataset.