Information retrieve is one of the most important operations in computer information systems. This paper presents a kind of fuzzy information retrieve method based on soft computing (SCFIR for short). SCFIR adopts fuzzy clustering analysis and artificial neural networks to organize databases in systems so as to increase the efficiency of fuzzy retrieve. At the same time, SCFTR realizes the understanding of query sentences in natural language by means of heuristic induct leading mechanism, interactional grammar analysis and semantic analysis. Finally, SCFTR applies several machine learning mechanisms (such as connective learning, evolutive learning, rough sets and so on) in the maintenance of system knowledge base, and thus the effectiveness and feasibility of retrieve systems are improved availably.