Recent experiments in oil shale processing, via in situ methods, have demonstrated the importance of proper bed preparation in obtaining high yields. While the creation of in situ retorts has been demonstrated to be viable, there does not exist a predictive design capability that can be used as a basis for optimization. A research program plan has been developed that delineates the tasks and costs necessary to develop a predictive capability for the rock fragmentation process. Models that treat fracturing, rock motion, permeability in rubbled media, and retorting are required to constitute a predictive capability. The plan indicates that these models must be developed in close association with a field experimental program. The field work is an evolving series of experiments from single borehole cratering tests to multi-level, multi-deck, mini-retorts. There is an optimum pace for the conduct of the program that requires some 7 years and $18 million for completion. Under current funding levels the work will require some 15 years and $27 million for completion. 16 figs., 10 tabs.