The disease management of traditional Chinese medicine (TCM) emphasizes identify patterns and determine treatment. Due to the symptoms of patients are usually untypical, the correctness of diagnosis has been criticized since the process of inferring symptoms to a particular syndrome is strongly related to doctor's subjectivity. The data mining methods, which effectively filter out attributes essential to TCM diagnosis from massive cases, provide a novel objective therapeutic guideline. In this research work, we explored the medical records of patients who had been visited TCM OPD for chronic cough during years 2004 to 2007. Cases of effective treatments were recorded into the database for data mining. The theory of decision tree was applied on the database to discover the implicit knowledge. The result suggested an objective diagnostic guideline. The correctness of this guideline was also evaluated.