Early corpus-based studies on genre analyses have shown many interesting empirical observations on linguistic variation across genres in the dimension of formality. Nevertheless, the intensively investigated linguistic features mainly involve lexical or grammatical level. Biber, D. (1988), for example, studied the similarities and differences among genres by focusing on 67 lexico-grammatical linguistic features. Since then, deeper level linguistic analyses, such as n-gram POS collocations, multi-word expressions and syntactic structures have been studied by many researchers (e.g. Fürnkranz, J., 1998; Zhang, W., et al., 2008; Moschitti, A., and Basili, R., 2004). Most recently, Fang and Cao (2015) involved semantic and pragmatic level genre analyses, as well as fine-grained linguistic features based on sophisticated annotation scheme. This paper mainly investigates the genre distribution of the six transitivity process types: material, mental, verbal, relational, behavioral and existential, based on M. A. K. Halliday’s Systematic Functional Linguistic Theory. A BNC sub-corpus containing eight genres is used as the linguistic resource, following the categories proposed by Lee, D. (2001). The transitivity process types of verbs are identified and annotated by a self-developed program which is rule-based. Detailed analyses on the different distributions of these transitivity types across the eight genres will be conducted, based on the identified empirical evidences. Meanwhile, automatic genre classification tasks will be implemented, aiming to testify the discriminative power of the transitivity types as document features, by feeding them to the state-of-the-art classifiers. The Natural Language Toolkit (Bird, S., et al., 2009) is used as the language processing and text classification platform. This paper assumes that the transitivity type will help improve the performance of the classifiers, by comparing the results of classification on using transitivity type, bag-of-words and other lexico-grammatical linguistic features as feature sets.