Institutional repositories (IRs) offer many clear benefits yet faculty authors have not demonstrated much interest in depositing their content into them. Without the content, IRs will not succeed, because institutions will sustain IRs for only so long without evidence of success. A yearlong study of faculty members at the University of Rochester has revealed some of the reasons why current IR systems are more useful in theory than in practice and has resulted in modifications to the University of Rochester's implementation of the DSpace code to better align the repository with the existing work practices of faculty. The findings have also caused a complete rethinking of how to explain and promote the IR. Research supported by a National Leadership Grant from the Institute for Museum and Library Services. (Contains 11 notes and 4 figures.)