Techniques that traditionally have been useful for retrieving same-domain analogies from small single-use knowledge bases, such as spreading activation and indexing on selected features, are inadequate for retrieving cross-domain analogies from large multi-use knowledge bases. In this paper, we describe Knowledge-Directed Spreading Activation (KDSA), a new method for retrieving analogies in a large semantic network. KDSA uses task-specific knowledge to guide a spreading activation search to a case or concept in memory that meets a desired similarity condition. Specifically, KDSA exploits evaluations of near-analogies encountered during the search to direct the search toward progressively more promising analogies. We describe a specific instantiation of this method for the task of innovative design, and we summarize the theoretical and experimental results used to validate KDSA.