In this paper, we introduce the inquiry optimization technique for a topic map database. We assumed a tolog-type query language and the topic map data structure defined by TMDM (ISO/IEC 13250 part-2). We show a strategy for making a retrival plan based on the typical size of topic map objects and the number of objects to be loaded in memory. As for implementation, we focused on one of the queries that has multiple plans, the type of query that retrieves topics related to a certain topic by some relationship both of which are specified in the query. We proposed an esimation formula for the retrival cost of each route and evaluated our technique by applying it to the topic map database called TOME, which is a prototype developed by the authors.