This chapter shows that treating the Chinese classifier system as a lexicalized semantic system based on shared ontology predicts both the agreement patterns that motivate the structure-based accounts, and the semantic selection patterns that motivate the cognition-based accounts. In the chapter, different perspectives toward classifiers are introduced including a cognition-based account (predicting a strong correlation with perception that is also robust and without exceptions but allows some fuzzy, overlapping classifications) and a structure-based account (predicting a strongly binary classification and a robust grammaticality judgement). Controversial issues regarding Chinese classifiers, such as the distinction between classifiers and measure words, the agreement between a classifier and its head noun, and the nature of 的 DE insertion, are explicated to show the pros and cons of various approaches. The authors demonstrate that Chinese classifiers are coherently organized in a ontology-driven lexical-semantic system. Major unresolved issues in the Mandarin classifiers system are closely examined at the end of the chapter.