Background: Herbal Medicine in Traditional Chinese Medicine (TCM) relies on interactions between the ingredients of a prescription. The combination is chosen to promote desirable interactions. Analyzing these interactions is an important step in quantitatively analyzing the effects of TCM on patient outcomes. The concept of interactions has not been adequately formulated in analyzing the effects of TCM due to the ambiguity of "interaction" and the need to go beyond traditional quantitative methods for analyzing data. Results: In this working paper, we present an exploratory analysis of clinical records of treatment using an interaction pattern mining approach. We present the most significant interactions found with a summary of the clinical significance of the interactions. Conclusions: Experimental evaluation confirms that this approach is able to detect effective high order herb-herb interactions in high dimensional TCM datasets. The interaction mining approach can be a potentially useful technique for discovering interactions not detected by other analysis techniques. The results will have implications for better understanding of the mechanisms of action and the overall system effects.