Peer-to-peer Web caching has attracted a great attention from the research community recently, and is one of the potential peer-topeer applications. In this paper, we systematically examine the three orthogonal dimensions to design a peer-to-peer Web caching system, including the caching algorithm, the document lookup algorithm, and the peer granularity. In addition to the traditional URL-based caching algorithm, we also evaluate the content-based caching algorithm for both dynamic and static Web content. Four different document lookup algorithms are compared and evaluated in the context of four different peer granularities, i.e., host level, organization level, building level, and centralized. A detailed simulation, using the traces collected at a medium size education institution, is conducted for the purpose of performance evaluation. Finally, several implications derived from this analysis are also discussed.