As computer and network technologies develop, information on line becomes flooded. Under this condition searching requested documents is really a difficult task. People are always seeking for an efficient searching tool and Topic Map is one of them. Topic Map is a new tool proposed by the International Organization for Standardization (ISO) to solve problems about knowledge representation and knowledge management. It has been widely used in many knowledge fields and in this paper we mainly make research on its application to information (document) retrieval. We firstly address some issues related to Topic Map: topic, association, occurrence as well as identity, facet and scope. By means of the current technologies for Topic Map developing, such as TMCL, TMQL, XML, SGML and so on, we propose a multi-layer Topic Map-based document retrieval model (TMDRM). TMRDM is based on Topic Map’s richly cross-linked structure and capabilities of topics used to group together objects that relate to a single abstract concept. This model helps people narrow the search scope step by step in order to facilitate them to proper documents.