Recognizing the need for analyzing large amounts of data in the study of online support communities, an automated content analysis method is introduced in this article. By adopting machine learning techniques and tools, this method requires minimal manual intervention while capable of analyzing large amounts of data automatically. Through this method, contents of messages from online support communities spanning over years are categorized as either informational support or emotional support. A case study on the analysis of online breast cancer and prostate cancer message boards is presented to demonstrate that the proposed method generates results comparable to results concluded from traditional manual qualitative content analysis methods.