Throughout the Plenary and Break-Out Sessions at the Workshop the issue was frequently brought up of linking together computational approaches which traditionally operate on distinct length scales. In response to this interest, an informal discussion was held to examine the prospects for a collaborative initiative on coupling atomistic and continuum methods. At this gathering everyone agreed that for materials modeling to be predictive it must be firmly based on realistic physical descriptions of the underlying processes, and that such realism should exist at all levels of the computational hierarchy. There was strong consensus that the integration of computational approaches across different levels (length scales) is a critical issue facing the materials modeling community. While the need to couple the atomistic and continuum methods is now quite widely recognized, few studies have been carried out to sufficient extent as to provide useful guidance for further work. All participants agreed that the next course of action should focus on specific problems. To acknowledge the fact that many participants have made contributions to a discussion session that did not follow any organized format, we have decided to use citations. However, the reader should be forewarned that our intention is to refer to contributions in the spirit of an informal discussion rather than any prepared and documented presentation. Under the circumstances, omissions and misinterpretations on our part are almost inevitable; for these we apologize. More accurate descriptions of the various views expressed on the issue of multiscale modeling can be found in the papers contained in Part III of these Proceedings.