The paper divides the corpus into six categories by analyzing annotated contents of the 31 overseas biomedical text corpuses and referring to UMLS semantic type according to the annotated entity types of the corpuses. It summarizes characteristics of the corpus in the aspects like semantic type and data source,expatiates on the building process and major steps of biomedical text corpus in the hope of laying down the foundation based on which related studies on China's biomedical text corpuses will be carried out.