As ubiquitous computing gained much attention in recent years, location estimation in wireless LAN becomes a hot topic. Previous research work suggests the use of the averaged Received Signal Strength (RSS) as fingerprint can achieve high accuracy for location estimation. In a library environment, however, the accuracy of such traditional approach is barely acceptable. It is because library contains considerably large number of metal bookshelves, and limited number of access points. Worse yet, the layout of these access points in the library is fixed for connection to the Internet, and therefore it is hard to change the environment to adapt for location estimation system. In this paper, we introduce an enhanced fingerprint (EFP) algorithm, and tested it in a library environment. The experiment result showed that the proposed EFP algorithm can have more than 30% of improvement in accuracy over traditional approaches without changing anything in the library environment.