As of today, there are scarce studies that show us how classifiers are used in real life. While information technology progresses, today’s researchers are privileged to use large-size databases to explore the real usage of classifiers. We selected 386 concrete nouns in Chinese, and looked them up in Chinese GigaWord for their collocating classifiers. Seventy individual classifiers were found to pair up with these 386 nouns. The following questions are of our particular interest: (1) What are the most frequent classifiers in Chinese? (2) On average, what is the number of classifiers paired up with a noun? (3) What is the number of nouns paired up with a classifier? (4) Which semantic dimensions are particularly relevant to the Chinese classifier system? In our data, a noun on average pairs up with 2.839 classifiers, and a classifier category generally contains 16 nouns. The most frequent classifier was zhi1 for animate beings. The noun di4tu2, taking 8 classifiers, is the most flexible noun in our corpus. By corpus data we will also talk about the role of ge0. Concerning the semantic diemsions, Tai (1994) identified four dimensions manifested in Chinese classifier system: animacy, shape, size, and consistency. And he also found a number of miscellaneous cases that reflect part-whole or metonymic relation. Our investigations show that some classifiers are not highly generalized, such as zun1 (for statues), chuang2 (for buildings), sao4 (for vehicles), etc. Those cases hint at the possibility that some classifier categories operate on low-level cassification such as VEHICLE and BUILDING, instead of highly-abstract semantic dimensions.