Based on the corpus-driven approach and the comprehensive study of the 207 Chinese separable words cognized in the Peking University corpus ”the General Outline of Vocabulary for HSK”, this paper draws a conclusion that 1. The separation degree of the Chinese separable words is not as high as expected and the joint forms of the separable words are ”regular” while the separated forms are ”idiosyncratic”; 2. The separating forms can generally be reduced to 13 categories. Statistics shows that the most frequent separating form is ”le”, secondly is complement, noun (pronoun), qualifier, and the least separating forms are verb components; 3. The separating particles and the numbers of the separable words with the separating forms form an inverse ratio: the fewer the particles of the separating forms are, the greater the number and higher the ratio the separable words have, and vice versa; 4. Most of the separable words with higher frequency are those used most frequently in daily life. The separable words with lowest frequency are those used least frequently in daily life.