Abstract - Data delivery to multiple recipients in networks can be achieved effectively via multicast based on a shared tree structure. This paper deals with a network-layer framework for efficient multicast through shared trees which are developed with an aid of guided information for locating nearest known on-tree nodes to connect upon receiving group join requests. Such information helps to shorten end-to-end delay over a multicast shared tree so developed, called a guided shared tree (GST). A designated node, known as a guidance information node (GIN), is employed to provide guidance for a group, and the GIN is found by joining nodes through the group ID indexing into a list of candidate GINs, in a way similar to automatic core discovery for CBT. The proposed multicast framework is evaluated by simulation and is shown to be efficient, readily suitable for large networks where group members account for a fraction of total nodes and are sparsely located.