A fundamental problem in XML query processing is tree pattern query (TPQ) matching which computes all data instances in an XML database that match an input TPQ. There is a lot of recent interest in applying materialized views to optimize TPQ matching. However, existing work in this area has focused predominantly on logical optimization issues, namely, view selection and query rewriting. With the exception of the recent work on InterJoin (which is primarily focused on path queries and views), there is very little work that has examined the important physical optimization issue of how to efficiently evaluate TPQs using materialized views. In particular, there are two related aspects of the problem that have yet to be adequately investigated: (1) what is an effective way to organize materialized TPQ views? and (2) how to exploit the organization of the materialized views to efficiently evaluate TPQs? In this paper, we present a new storage scheme for materialized TPQ views and a novel eva...