The indexing problem, the problem of how to label cases in memory and retrieve them as needed, is a central issue in all case-based reasoning (CBR) systems. Most work on the indexing problem to date has been completely process oriented, that is, it has focussed on algorithms that can be used to retrieve cases quickly. Almost none of this work has looked at the content, or meaning , of the indices used to index cases. This paper is a rst attempt to redress that imbalance; it presents a content-driven approach to indexing and reports preliminary results from our working group at Northwestern University's Institute for the Learning Sciences (ILS). This group's goal has been to devise a content theory for indices of suucient generality to support the full range of CBR projects at ILS. 1 Most projects at ILS involve retrieval of stories about intelligent agents and their actions; stories are cases rich in casual and intentional connections amongst agents their actions and resulting states. Accordingly, we have tried to construct an index format capable of supporting remindings between stories based on a rich intentional vocabulary. We believe this will prove to be the right vocabulary for supporting all cross-contextual reminding. Our immediate objective however is a pragmatic one. We have not set ourselves the goal of discovering the \right" set of indices, that is, one that is valid in any absolute theoretical sense. We are in search of a set of indices that is suucient for our purposes.