Abstract-We present an approach of selecting medical literature for the purpose of supporting clinical diagnostic decision. The proposed approach is based on link analysis in the weighted keywords network. The keywords in the network are extracted from the keywords lists in the articles, the Mesh lexicon and acronyms, with which the full relation of keywords and articles (the link network) is built. In this paper we introduced in details how to extract the collection of valid sentences and diagnosis-related keywords based on the keywords co-occurrence analysis and existing description of symptoms. The paper also lays out the detailed process of measuring the relevance between keywords and articles with link analysis in the weighted keywords network. Finally, we report some related experiments and the results of user evaluations.Index Terms-Clinical decision support, link analysis, weighted keyword network, medical information retrieval.