Grafting an ATM network onto an existing Ethernet network is a non-trivial exercise in network configuration. The process of mating ATM technology to an Ethernet network without tampering the underlying backbone and network configuration gives rise to a variety of networking issues including addressing, compatibility, security, efficiency, administrative effort, and scalability. Examples of actual situations serve to illustrate these issues. We have designed several scenarios to tackle the topological problems of an ATM graft. A dual IP approach works but suffers from a lack of scalability. We expect an edge routing solution, with subnetting, will be the most elegant and scalable.