We describe our participation in the INEX 2009 Entity Ranking track. We employ a probabilistic retrieval model for entity search in which term-based and category-based representations of queries and entities are effectively integrated. We demonstrate that our approach achieves state-of-the-art performance on both the entity ranking and list completion tasks.