BACKGROUND: The amount of health data accessible on the Web is increasing and Internet has become a major source of health information. Many tools and search engines are available but medical information retrieval remains difficult for both the health professional and the patients. OBJECTIVE: In this paper we describe heuristics that aim at matching as much as possible queries with the content of the documents in the context of the CISMeF catalogue (Catalogue and Index of Health Resources in French) and its Doc'CISMeF search tool. The queries are represented by terms and the content of the documents is indexed by a terminology based on the MeSH thesaurus. RESULTS: Several operations are performed to match the terms of the terminology: natural language processing techniques on multi-words queries, phonemisation, spelling correction, plain text search with adjacency etc.. Each one is tested to evaluate its contribution in matching the terminology and the indexed documents. CONCLUSION: The implemented heuristics contribute significantly with good results in maximising as much as possible the recall of the Doc'CISMeF search tool.