The primary LAN technologies in use today --- ethernet, FDDI, and token-ring --- all provide hardware support for broadcast and multicast capability. However, distributed systems have traditionally used unicast messaging exclusively, even when multicast communication patterns arise. As the number of distributed applications grows, the load on networks caused by unnecessary unicasts will increase. In addition, for some applications performance and group size are limited by using unicast technologies for multicast. If multicast technologies are exploited, the network load caused by "redundant" packets will be reduced. Exploiting multicast will also improve the performance and scalability of some distributed applications. However, as distributed systems move towards exploiting multicast, multicast flow control protocols are becoming more important. Finding a general, effective solution for multicast flow control will help facilitate the exploitation of the multicast primitives provided in LANs and Wide Area Networks (WANs). This dissertation presents the results of an investigation into multicast performance on local area networks. An analysis of multicast flow control is presented which distinguishes between {\it rate reservation} and {\it rate control}, followed by a discussion of the major design issues associated with multicast flow control and a presentation of a proposed protocol family. The proposed protocol family is based on send-rate control. An unreliable and a reliable multicast flow control protocol based on the proposed family are presented. A study of the performance of these protocols is also presented. This dissertation concludes with an investigation into how well the proposed reliable multicast flow control protocol performs when used to disseminate messages. This dissertation argues that direct rate control has merit as a technique for multicast flow control on local area networks.